CONSTITUENCIES AND ROLL CALL VOTING: AN ANALYSIS OF TILE HOUSE OF REPRESENTATIVES FOR THE 88TH CONGRESS Thais for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY JACK RONALD VANDERSLIK 1967 LI BRA R Y MiCI'igan Sta to U IVCE‘Sity -1". rt- This is to certify that the thesis entitled Constituencies in Roll Call Voting: An Analysis of the House or Representatives for the 88th Congress presented by Jack R. VanDerSlik has been accepted towards fulfillment of the requirements for Ph.D. degree in 2011 tical Science Major professor Date | ABSTRACT CONSTITUENCIES AND ROLL CALL VOTING: AN ANALYSIS OF THE HOUSE OF REPRESENTATIVES FOR THE 88TH CONGRESS by Jack Ronald Vanderslik This study explores the relationships between selected constit- uency variables and roll call voting patterns of representatives elected from these constituencies. Several normative and empirical reasons sug- gest that representatives will accommodate themselves to constituency pressures in their voting. This study inquires whether these pressures are discoverable as relationships of economic, social and demographic characteristics to measures of representatives' voting. This study examines the House of Representatives of the 88th Con- gress (elected in I962). The data include roll call voting records of all representatives and twenty-one constituency variables for each con- stituency. The latter, drawn from I960 census data, are: (I) percent population change, (2) percent urban, (3) percent Negro, (4) percent foreign stock, (5) median age, (6) percent of elementary enrollment in private schools, (7) Percent of population with less than five years schooling, (8) percent of the population with high school, (9) percent of population with college, (10) median education, (ll) median family income, (12) percent unemployed, (l3) percent owner occupancy, (IA) per- cent of sound housing units with all plumbing, (IS) median rooms per Jack Ronald Vanderslik housing unit, (16) median persons per occupied unit, (17) median home value, (18) median rent, (19) percent white collar workers, (20) percent blue collar workers, (21) percent farmers. The roll call voting data are organized, by (Guttman) cumulative scaling technique, into sixteen dimen- sions of voting. These scales are Foreign policy, Foreign trade, Consu- mer protection, Conservation, Agricultural policy, Urban improvement, Negro rights, Civil liberties, Aid to education, Orientation to profes- sionalism, Labor versus business, Spending, Space spending, Orientation to [government] debt, Social welfare, and Party loyalty. With certain exceptions, l scaled and scored all members of the House according to their voting on each dimension. The statistical tools used are mainly correlation and analysis of variance. Rank correlation coefficients are produced for every constituency variable with every roll call voting dimension. To evalu- ate the correlations I defined weak relationships as those with correla- tion coefficients lower than t .250. ”Strong” correlations are those equal to E .400 or higher. I'Moderate” relationships are those between t .250 and .399. I hypothesized a complete description of relation- ships between the constituency variables and the scale scores of voting, and I reported the findings in similar fashion. The correspondence be- tween the hypothesized and observed relationships is low. Rather than each constituency variable being correlated strongly with one or two dimensions, a few constituency variables are strongly or moderately cor- related with several dimensions and the rest of the relationships are weak. The constituency variables most clearly related to policy voting are (2) percent urban, (4) percent foreign stock, (7) percent with low Jack Ronald Vanderslik education, (14) percent owner occupancy, and (21) percent farmers. Sur- prisingly low relationships resulted from correlations with (9) college education, (11) median income (19) white collar, and (20) blue collar. Partisanship is much more strongly related to roll call voting than any constituency variable. 0n fourteen dimensions the correlations with party range from .490 to .821. Half are .700 or higher. Within parties, constituency characteristics show greater relationships to variations in Democratic voting than Republican voting. This is ex- plained by the sharper differences between the southern and nonsouthern wings of the Democratic party compared to the eastern and western divi- sions of the Republicans. Democrats from electorally competitive districts vote more liberally than Democrats from less competitive districts. Republicans from competitive districts voted more conservatively than those from less competitive districts. Ambitious representatives voted more liberally than their less ambitious colleagues. Reelected Democrats voted in no systematically different way than those who left office, but reelected Republicans voted more liberally on social policy dimen- sions. CONSTITUENCIES AND ROLL CALL VOTING: AN ANALYSIS OF THE HOUSE OF REPRESENTATIVES FOR THE 88TH CONGRESS By Jack Ronald Vanderslik A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1967 fl? '\:Copyright by JACK RONALD VANDERSLIK 1968 ACKNOWLEDGEMENTS During the course of this study I have had help from many people. Robert Scigliano encouraged my interest in the relationships between constituencies and representatives. My dissertation committee members, Joseph Schlesinger, Harold Spaeth, and especially my chairman, Charles Press, were prompt and helpful critics of the manuscript. Professor Press rescued the entire project by taking on the committee chairman- ship at a late date. Professor Spaeth made telling improvements on an earlier draft of the second chapter. Several fellow graduate students helped me unmuddle my thinking at various stages of the project. 1 par- ticularly want to thank Joe M. Allman, Eric Carlson, and William 0. Ice for the many hours of discussion they spent with me. Michigan State University provided several thousands of dollars worth of computer ser- vices without which the study would have been impossible. I am most indebted to my family and my wife, who encouraged my work. Together we thank our heavenly Father for meeting all of our needs. Jack Ronald Vanderslik June 26, 1967 East Lansing, Michigan TABLE OF CONTENTS ACKNOWLEDGEMENTS . LIST OF TABLES Chapter VI. APPENDICES . BIBLIOGRAPHY . A THEORETICAL OVERVIEW . The Scope and Rationale for this Study Review of Related Research Data and Tools of analysis ROLL CALL VOTING ANALYSIS: THEORY AND PROCEDURES Scaling Roll Call Voting in the 88th Congress Scaling Results Hypothesized Relationships Between Constituency Characteristics and Scale scores THE RELATIONSHIPS BETWEEN CONSTITUENCY CHARACTER- ISTICS AND ROLL CALL VOTING. . . . . . . Correlations Between Independent and Dependent ° Variables Statistical Controls and Limitations Summary RARTY VOTING IN CONGRESS AND THE RELATIONSHIPS WITH CONSTITUENCY CHARACTERISTICS. Findings Summary INTRA-PARTY PATTERNS IN ROLL CALL VOTING . Competitiveness Ambition Reelection Summary SUMMARY AND CONCLUSIONS. Summary of Findings Apologies and Caveats Page 42 99 IA] 180 215 230 279 Table I-l. 11-2. 11-3. Ill-l. III-2. III-3. III-4. III-S. IV-l. IV-2. IV-3. IV A. LIST OF TABLES Item and Measures for Congressional District Populations Variables. The Scalability and Distribution Characteristics of the Sixteen Roll Call Voting Scale Scores Intercorrelations of the Economic Variables Hypothesized Relationships Between 21 Independent Constit- uency Variables and 16 Roll Call Voting Variables Rho Correlations Between 21 Independent Constituency Variables and 16 Roll Call Variables . Observed Relationships Between 21 Independent Constituency Variables and 16 Roll Call Voting Variables. Direction and Strength of the Correlations Between 21 Independent Variables and 16 Roll Call Voting Dimensions . Demonstrating Nonmonotonic Relationships Between Median Family Income and Roll Call Voting Variables . Rho Correlations Between Representatives' Party and 16 Roll Call Voting Variables . The Degree of Differentiation Between Republicans and Democrats on the 16 Dimensions of Roll Call Voting (Kruskal-Wallis H Coefficients). . . The Distribution Characteristics of the Scale Scores on Sixteen Roll Call Voting Dimensions: By Party . Comparison of Constituency Characteristics by Parties: 257 Democratic Districts and 178 Republican Districts of the 88th Congress . . . . . . . . . Strength of Observed Relationships Between 21 Indepen- dent Constituency Variables and 16 Roll Call Voting Variables. Each Party is Correlated Separately Page 37 38 83 89 95 96 10h 105 106 108 109 Ill 133 IAO lh6 151 154 157 158 Table lV-S. IV-6. V-l. V-2. V-3. V-h. V-S. V-6. V-7. V-B. Direction and Strength of the Correlations Between 21 Independent Variables and 16 Roll Call Voting Dimensions Each Party is Correlated Separately. Ranks of the Rho Correlations for Six Independent Variables Over the 16 Dimensions of Roll Call Voting . The Degree of Differentiation Between Southern and Nonsouthern Democrats on 16 Dimensions of Roll Call Voting (Kruskal-Wallis H Coefficients) . . . Comparison of Constituency Characteristics Within Demo- cratic Districts by Regions 89 Southern Districts and 168 Nonsouthern Districts of the 88th Congress . The Degree of Differentiation Between Eastern and Noneastern Republicans on 16 Dimensions of Roll Call Voting (KruskaI-Wallis H Coefficients) . . Comparison of Constituency Characteristics Within Repub- lican Districts by Regions SA Eastern Districts and 124 Noneastern Districts of the 88th Congress The Degree of Differentiation Between Democratic Repre- sentatives from More and Less Competitive Districts on 16 Dimensions of Roll Call Voting (KruskaI-Wallis H Coefficients). . . . . . . . . . . . . . . . . . . The Degree of Differentiation Between Republican Repre- sentatives from more and less Competitive Districts on 16 Dimensions of Roll Call Voting (KruskaI-Wallis H Coefficients). . . . . . . . . . . . . . . . . The Impact of Progressive Ambitions on Roll Call Voting (KruskaI-Wallis H Coefficients for each Party Separately . Comparison of Reelected and Nonreelected Representatives on 16 Dimensions of Roll Call Voting (KruskaI-Wallis H Coefficients for each Party Separately). . . . vi Page 159 160 171 172 18h 186 I91 I93 200 203 208 211 CHAPTER I A THEORETICAL OVERVIEW Most often students of legislative behavior have examined the parties which organize the activity and compete for control in the legislative arena. Several analysts have examined the importance of legislative parties on roll call voting (e.g., Lowell, 1902; Turner, 1951; Jewell, I955; Truman, 1959). Legislative bodies have been charac- terized according to the degree of partisan voting in them, and legis- lative parties are described in terms of their cohesiveness. The behavior of individual legislators has been compared to the ”normal” or ”average” behavior of their fellow partisans. When low partisanship ' occurs variables such as district competitiveness, seniority, ideo- logical commitment, friendship patterns, and group memberships have been suggested as explanations for this finding.- In some cases the degree of inter-party conflict in roll calls varies according to the nature of the POIicies under consideration. For example, welfare issues have led to sharp cleavage between the parties in Congress, but proposals for veterans' benefits have not (Turner, 1951, p. 70). Other scholars of legislative behavior have studied the nature of legislative constituencies. It is generally acknowledged that a legislator's behavior is partly accounted for by complexion of the con- stituency from which he comes. However, the relationship ought to be I stated in more precise terms. Under what conditions is there a rela- tionship between constituency characteristics and the voting of the representative from that constituency? The present study focuses on this problem. The Scope and Rationale for this Study The setting in which this question will be examined is the House of Representatives. The members and districts are appropriate and con- venient to study for several reasons. Much of the business of the House-- its hearings, debates, and roll call votes--is a matter of public record. Most representatives participate in the roll call votes. Also most rep- resentatives are elected from single member districts. Systematic data are available for each of these districts. Additional advantages will be apparent as this study develops. My analysis is limited to the 88th Congress which was elected in November, 1962, and held office in 1963 and l96h. This is a short period of time to use as a basis for generalizing about legislative be- havior. A longer period would have some obvious advantages. However, there are difficulties implicit in longitudinal studies. If several con- gressional terms are used, several important relationships, both in and OUtside the House, are subject to change. There may be alternation of Party control in the House. This means change in the leadership of the House as a whole as well as in the committees. Over time the membership changes even when there are not alternations of party control. There are changes in the districts too. This is not limited to the term following a decenial census. There may be turnover in the party controlling the office of the President. These and additional changes make it difficult to interpret differences between the nature and implications of decisions taken withinany substantial period of the history of the House of Rep- resentatives. It seems worthwhile, therefore, to proceed with an ex- ploratory study of a single Congress, exploiting the advantages which such a study has while keeping in mind the limitations which it imposes. The theoretical basis of this study assumes relationships de- scribed in the concept of representation. In most processes of gover- ning, men in authority positions have claimed to represent, speak in behalf of, or make decisions in the best interests of those who accept them as authorities. This indicates that lines of responsibility in the relationships of representatives have often been vague. Those claiming to be representatives have often preferred only vague obligations to the constituents who hold the representatives in esteem. These arrangements are what Alfred de Grazia (1951, pp. 1349) refers to as "virtual rep- resentation" and are consonant with the views of Edmund Burke, who em- _ Phasized the oneness of the national interest and the oneness of the deliberative body of the nation. There are two aspects to representation by which A, the repre- sentative; and B, the represented; are associated. The first is that A has authority to make decisions binding upon B. B is expected to ful- fill the obligationsA makes in B's behalf. Virtual representation does "0t 90 beyond this aspect of relationship. However, the second aspect makes A responsible to B. This means that A is subject to certain sanctions by B which 8 may use if A im- poses Obligations which are not to 3'5 liking. According to Beard and Lewis (1932), the means for enforcing the second aspect is implied in the fact that the representative is elected by those for whom he is to act. A' s authority is formally defined, and his right to exercise that authority is conditional upon approval by B. The knowledge that A is responsible to B is not sufficient for predicting how A will carry out that responsibility. Normative theorists have suggested at least three ways in which this responsibility can be car-r ied out. First is the view that the representative will be acting in the best interests of his constituents when he is free to follow the dictates of his reason and conscience. The second view requires repre- sen tatives to be true to the policies of the party because he was elected in a partisan election. The third view requires that considerations of conscience and party must always be subordinate to the will of the con- sti tuents. Each of these views requires some examination, but it should be POi nted out that there is no necessary reason why conflict will always exist among these prescriptions. The three objects of responsible be?- haVIOr may accept the same decisions by the representative; that is, a rePi‘esentative may vote "yes" on a civil rights proposal to the satis- 1:afition of his conscience, his party, and his constituents. Nevertheless the way in which the representative would rank these remains relevant: N If the interests of conscience, party, and constituency are in conflict, “hlch one will the representative follow?" Recent literature provides some insights, but not many firm conclusions about which of these objects representatives regard most highly. It is difficult to examine the extent to which representatives follow their own inner direction. It is hard to know what criteria the inner directed legislator uses and, without a knowledge of these criteria, impossible to assess whether the representatives vote according to the di rectives of their consciences. Normative theorists have suggested a wi de range of standards for the representative to regard. They might be eternal verities, such as Socrates' concept of justice, or they may re- flect Burke's (originally published in 1790; republished in 1955, p- 265) view that representatives meet together so that the government can ,act as a trustee for the whole nation. Wahlke, gal. (1962) used the term “trustee" to refer to state legislators who said their, style of representation was that of making decisions on the basis of their own convictions and principles. Their interview data revealed that most of the respondents who could be classified (161 out of 21+l)l identified with the trustee style. It was not determined whether or how often these trustees make roll call voting decisions which they perceive to be con- 515tent with their consciences but are contrary to party, constituency, 0" both. It is probable that the frequency of such conflict is rather IOW- Usually representatives have learned their political values in the comn'Il-lnity they represent. With a modicum of competition in the recruit- ment process, it is unlikely that candidates with views highly unpopular ‘“ their districts will win election to office. Thus representatives may per‘Czeive themselves as following the dictates of their own consciences, "I'll le in fact they are accommodating themselves to views and attitudes that are held by their constituents. Legislators may identify with the truStee role as a matter of lip service to the cult of rugged indivial- ism. I am skeptical of the inference that representatives frequently This seems to be an embarassingly low response level when one realizes that h7h legislators were interviewed. The authors do not account for or speculate on the effects that these missing data may have had on the findings‘reported. chii‘B contrary to their perceptions of the interests which are important 3r, t:f1eir constituencies. It may be that the trustee role is a comfor- tab 1 e identification for representatives who come from constituencies \NPIi czlfi contain competing interests (See Dye, 1965, p. 182). The repre- sentative might see himself as acting on his own, while his choice re- flects his internalized conception of how constituency alignments are balanced. The view that representatives should vote for the party position is. ear1 element of the argument for responsible parties (see Schattschneider, 1942, and the Committee of Political Parties of the American Political Science Association, I950). This argument implies, with Burke, that a national focus is better than a local one, and that disciplined parties cart vvi thstand the narrow demands of self interest that come from small single member districts. Turner (1951) reduced the force of S"-"l'lalitschneider's argument by demonstrating that fellow partisans do vote with significant unity in the House of Representatives. Others (To? example, Ranney and Kendall, 1956) have advanced the view that the pol itical stability of the nation has been enhanced by the absence of Intensely ide0109ical parties. Concerning Turner's findings more will be Said later. The fact that there is a strong measure of party unity may, indicate that representatives consider that party loyalty is desir- able; and necessary. It has not been shown, however, that partisan unity IS Independent of constituency characteristics; much less that party POSltions are the basis of decision making by representatives in op- POSition to the interests they perceive in their constituencies. A commonly held view is that representatives ought to vote as tr1eei'r constituents wish.2 For some constituents this view constitutes 3,1 . i nformal rule of the game. Because the constituency electorate is trjez final judge of whether the behavior of the representative has been ,aF>F>r'opriate, this rule can be enforced. Whether electorates do in fact esgeer'cise such discrimination is not certain, but clearly the opportunity ‘fc>r' it is undeniable. Further comment on voting by constituents will bee ggiven later. It has not been shown that constituents approve of attempts by the political parties to sanction representatives who have been irregular in their party support. The legislative parties have some sanctions at their disposal but their effectiveness in determining the decisions of individual representatives is not clear. Occasionally, they may be important considerations, other times they may be of little Significance. They are seldom very public and rarely do they cause the defeat of an incumbent. Elections provide the circumstances in which constituents can forma 1 1y decide whom they want to represent them. There are l+35 congressional seats in the House of Representatives. All citizens \ 2 Cantril (1951, p. 133) reports three surveys which asked citi- Zens how they thought Congressmen should vote. The way Use their Don't know, people feel own judgment no opinion "Member 1938 37.Li% 51+.l% 8.5% August 1939 61 39 April 191m 66 34 April 1940 63 37 Similar findings come from Public Opinion Quarterly X (Summer, 1946) pp. 268-269. NORC, April 28, 191-16 except for those in Washington, D.C. are constituents of at least one representative and most are constituents in single member districts.3 It i 5 important to note that each representative must submit to the e].g=;.:-_:toral process frequently. Every two years there is what Key (1964, 5.1.3.13- ed., p. Lil-19) refers to as the "double jeopardy" of contests for nomi nation and election. The geographic boundaries by which constituen- ci es are legally defined are not frequently changed. The populations with i n most districts are relatively stable. Because incumbent repre- sentatives are eligible for reelection, it is advantageous for them to establish enduring bases of electoral support among their constituents. The partisan context of congressional elections cannot be ig- nored . Stokesand Miller (1962) have shown that the vast majority of voters are party identifiers who support their own party consistently, Particularly in congressional contests. Some districts are clearly safer than others (Cox, 1962; Wolfinger and Heifetz, 1965). It is necessary to note that there are no congressional elections 3g; _s_e. There are often $19"1i‘Ficant interaction or coattail effects (Moos, 1952; Press, 1958; Meyer’ 1962, pp. 52-61}; Key, 196‘}, 5th ed., pp. 556-574) between candi- date-S for the House and those for other national, state, and local \ c"mgct‘essmen should rely on Congressmen do rely on . Ountry l43% Country 9% Consti tuents 39 Constituents 40 Own judgment 18 Own judgment 51 TheSe are not highly consistent results. Certainly the format in which the questions are asked affect the responses. However, it is clear that a large proportion of people believe constituency preferences ought not be ignored. 3A community member may have more than one representative if con- gressmen are'elected at large. For the'88t'h Congress, elected in 1962, Hawaii and New Mexico elected delegatiOns of two members at large. Con- necticut, Maryland, Michigan, Ohio, and Texas elected one representative Of-‘f'i ces on the same party ticket. These considerations make it obvious the t the contest for office in any particular House district cannot be thought of as an independent event. The issues are shaped by the parti- sari balance within the district (Ennis, 1962) as well as in the state and nat i on. The congressional district is an enclave of the state and na- ti on and the salient issues of the campaign are sometimes defined by the. party nuclei at these higher levels (Schlesinger, 1965). The party con- 1 text within which recruitment to congressional office occurs may result in differences between representatives in their style of representing their orientations to ideological issues, their party regularity in roll cal l voting, their desire for personal voter support, and their age (Snowi 55, 1966) . The partisan context of congressional elections is not itself an '"dependent variable. Partisan identification has many correlates, if "012 causes, among socio-economic and demographic variables. Although there are marked differences between the conceptualizations, methods, and inferences of Lazarsfeld and his colleagues (191.4 and 1948; 1951+) on one hand and the Survey ResearchCenter teams (1951+; 1960) on the other, both show that socio-economic and demographic variables differentiate be- tween both party voters and party identifiers. A central proposition of 1-.Le&ople's Choice was the view that "Social characteristics determine \ at large while the other delegation members were from single member dis- trlCts. Alabama elected eight representatives at large. Its delegation was reduced from nine to eight by reapportionment. Under Alabama's unique 9-8 plan each of the old nine districts individually nominated its 1"CUmbent representative in a primary. A statewide Democratic runoff pri- nary was held and the low man was eliminated from the Democratic slate. The eight then ran at large in the general election and all were elected. (ELAnnuaI Report, XX, 1961+, p. 378) lO ID‘D'I itical preferences” (1944 and l9h8, pp. 27, I38, 148).4 This was re- asserted in m. "Socioeconomic status . . . is directly related to trjea final vote decision (Chart XX). The higher the socioeconomic status (SES), the more Republican the vote; put crudely, richer people vote Re- pmJta‘lican more than poorer people'I (1954, pp. 55, 56). In The Voter De- <;i (ices, Campbell, Gurin and Miller relate voting to a list of ten factors: 1. Type of community (size of community in which voters live) Socio-economic groups to which voters belong (social and economic status) N 0 Occupation Trade-union affiliation Income 3. Ethnic origins of voters A. Religious affiliation 5. Age 6. Sex 7. Education 8. 9. 0. Later- Campbell, _til” (1960, especially pp. 156-159) demonstrated clear differences in party identification in their comparison of working class PeOple to members of the middle class. Though the contrast in regions is apparent, the class component <>f American politics is also clearly seen. In the South, where Demo- czratic adherence was for many years an essential feature of regional l<3yalty, both classes are heavily Democratic although the middle <:1ass dilutes its support of the Democratic Party somewhat by a ten- |‘.>olitan area as defined by the United States Census of I930 or ISAC" (p. 7%). All others were defined as rural districts. Unfor- tuna tely most of the applications of this tool were done in relation to par ty - The perhaps unfounded assumption was that the difference by met- ropo I itanism alone might be spurious because party would then be con- f()L"-"J assume that the participants want to support a relatively consistent and fixed substantive position, then the roll call records for each of the motions (items) should correlate positively with one another. If they do not correlate, they are probably measuring different kinds of support, not the same kind. The correlation test bolsters the intuitive test that the items of the scale measure the same thing. Besides including inappropriate items, like labor union support scores by two representatives may not mean the same thing. Knowing that ranesentatives A and B have registered 20 percent support for unions does not tell us on which items each has registered support. Each voted in support on one-fifth of the proposals, but all of the items endorsed bYA may have been opposed by B, and vice versa. This characteristic handicaps analysis. Guttman scaling provides means for overcoming some of these dif- ficulties. Originally developed to determine whether a series of 45 interview questions tapped an underlying attitude, the scale dimensions revealed by this technique have sometimes been treated as attitudes (Schubert, I959, p. 273; I965, pp. 27, I9l ff. See also the criticisms and reformulation by Spaeth and Peterson, I967). This confusion has its roots in the early presentation of the theory and application of scale analysis by Louis Guttman (Stouffer, _£._l-: I950, pp. 3-90). Stouffer notes that “Guttman's model deals only with the manifest relationship among attitude items and defines an attitude directly as the observed responses to those items” (pp. 6-7). Guttman suggests that a definition might be derived if one assumes that an attitude is a ”delimited total- ity of behavior with respect to something'l (p. 5l). For Guttman this delimited totality implies that there is a universe of behavior attri- butes which can be sampled by means of questionnaire items. The ques- tionnaire responses allow the analyst to determine whether there is a rank order of attributes--a scale; and if there is, the rank of the in- dividual respondents on the dimension the scale represents. Scales are relative to time and to populations, allowing the inference that an atti- tude is not a persisting organization of beliefs. I prefer to treat Guttman scaling as a technique for assessing the dimensionality in decisional responses of representatives. In this way the unnecessary freight implied in the conflicting definitions and understandings of the concept attitude does not weight down the proposed study. Spaeth and Peterson argue that: . . we cannot assume that the rank order on voting behavior re- veals the rank order on an underlying attitude. A cumulative pattern may be considered a necessary but not a sufficient criterion for the identification of an attitude. That is, a single underlying attitude implies a cumulative pattern, but a cumulative pattern does not imply a single underlying attitude. 46 If we are told that there is a cumulative pattern of behavior, we do not know whether a single attitude or a set of systematically related attitudes generated the pattern. If the latter is the case, one explains nothing by referring to the scale derived from votes as an attitude (p. l4, prepublication draft). This is consistent with Torgerson's (I958, pp. 298-345) inquiry into the usefulness of Guttman scaling. The primary problem will be to determine whether the set of items and the set of individuals together ”form a scale.” Can we order the subjects and/or the items along a continuum in such a way that the responses of subjects to items can be accounted for by this or- der? Can we consider the responses to the items to be dependent on a single (though perhaps complex) attribute? Can the responses to the items be considered as indicating a relationship, with respect to a single attribute, between the ”position'I of the subject and that of the item categories? Can the alternative response cate- gories within an item be ordered? If so, we shall say that the items form a scale for this group of subjects, or that the attri- bute to which all of the items refer is a scalable attribute (pp. 301-302). The meaning of the attribute whose dimensionality is being tested . is specified by the analyst and is operationalized in the items he uses._ In Guttman's description the items used are a sample from a universe of content.2 However, this is a hypothetical universe. Guttman's (Stouf- t al., I950, p. 8l) ”universe of all possiblegguestions of the same fer, content” cannot be constructed or tested. The theoretical difficulties of this idea have been noted by Torgerson (I958, pp. 332-333 and passim). It remains for the analyst to devise or select the questions which he believes will tap a particular attribute in the respondents. That is, the analyst assumes the existence of the attribute and that respondents can be ordered with respect to the attribute on one continuum. Several 2Guttman uses the term I'universe of attributes.” l have avoided 'the word attribute because I will use it as Torgerson (I958) does in the Fusrtion already cited. In Torgerson's usage llunivese of content” and "universe of items” are used interchangeably and his use of the term item "corresponds closely” to Guttman's term attribute (Torgerson I958, r>- 331). L3 . e c I I e,» \rs .5- n.- p \ I; '5 1" l (K> 47 fixed response items will allow the analyst to estimate the respon- dents' positions on the continuum. Items may have multiple response categories or may be dichotomous. Assuming that the respondents are distributed along the attribute continuum, the more broadly the respon- ses are distributed, the more discrimination and information the analyst can obtain from the scale. The existence of the hypothesized scale, developed from an as- sumed underlying attribute, can be tested. If all the items do inquire reliably about the same attribute, and if all the respondents correctly perceive their own relationship to the attribute and can identify their relationship to it on the items, then the analyst should have a perfect scale. Perfect scales are rarely reported.3 It is more common to re- port a response pattern which approximates a perfect scale closely enough to be treated as if it were perfect. Several criteria have been suggested for evaluating whether the data obtained sufficiently resemble the perfect scale which is hypothesized (Torgerson, I958, pp. 322-324). The main one is the Coefficient of Reproducibility. CR = I - %- .1 re- fers to the total number of non-scale responses and n_is the total number of responses. A CR lower than .90 means that the data do not support the inference of a single underlying attribute dimension. Torger- son supplements this criterion with several others. ltems with 3Schubert (I965, p. 99) records that early in his application of scale analysis he made a ”lucky hit,” finding that a subsample of eight "right to council” decisions by the Supreme Court for the period I940- l947 scaled without a single inconsistent vote. ”The reason it was lucky," he says, ”is because the observation of this rare empirical event lencouraged me to look further into the structuring of judicial attitudes; in dozens of cumulative scales of the voting behavior of Supreme Court justices and of other judges, which I have constructed since that tzime, I have yet to encounter another perfect scale.” 48 extreme marginals, i.e., those on which more than 80 percent of respon- dents give a like response, should not be used in determining the CR. Scales of dichotomous items should include IO items or more. No large proportion of the respondents should have the same pattern of inconsis- tent responses. Each item of the scale should have a CR of at least .85. Each item category should have more consistent than inconsistent responses. These are convenient criteria, but not necessary ones. I will specify the extent to which I have followed them in the scales of legislative voting. The advantage of the Guttman scaling technique is that it yields a ranking with a known percentage of non-scale responses. That is, with knowledge of each respondent's rank on a scale, the entire response pat- tern can be reproduced with a known proportion of errors (I -— CR). Further, there is a built-in criterion for rejecting items which return responses that are inconsistent with the larger pattern. The analyst uses a number of items to tap the hypothesized underlying attribute. The marginal total of yeas and nays determines the rank of the items in the array. But if many respondents do not have transitive response pat- terns and if the non-scale responses occur mostly on a particular item, that item should be removed from the scale. The scaling procedure pre- vents the analyst from including inappropriate items, items which are not part of the universe of content for the attribute under examination. Because scales yield ranks of respondents on attributes, these ranks can be compared to other variables, taking advantage of all the properties eassociated with ordinal level measurement (Siegel, I956, pp. 2l-30). 49 Scaling Roll Call Voting in the 88th Congress To hypothesize dimensions of voting in the 88th Congress, I re- viewed all the roll call votes of both sessions. The basic sources of information were Volumes XIX and XX of the Conqressional Quarterly Alma- 135; (l964 and I965) and the CQ Weekly Reports for the same period. ll9 roll call votes were reported for the House during the first session and in the second session II3 were added to the record. I examined these votes with a previous knowledge of dimensions defined by earlier analy- ses of House voting--Farris (I958), MacRae (I958 and I965), and Riesel- bach (l964 and I966). | roughly defined eleven policy areas which might yield dimensions: A. Foreign policy B. Consumer protection and conservation C. Agricultural policy 0. Urban improvement E. Negro rights F. Civil liberties G. Education improvement H. Orientation to professionalism l. Economic policy J. Social welfare K. Party loyalty _ Several residual categories were named. Some items were expected to be unique and some ambiguous (or multidimensional). I expected some votes to be in response to particular institutional actors; for example, the .President and the Senate. Some issues might pit the government against business and labor. In some cases the House might divide with the lead- ership of both parties opposed to dissidents of one or both parties. The first sorting using the above categories made several things clear (see Appendix A). Few roll calls seemed to fall into the residual categories other than “unique” and ”ambiguous.' Seven suggest the 50 leadership versus a few dissidents, but on only one item do the dissi- dents constitute nearly I0 percent of the participants. It would be impossible to assess the validity of a scale based on such items because any CR derived would obviously capitalize on the low proportion of oppo- sition votes. Twenty-two items were dismissed as ambiguous for one rea- son or another. A couple of items had a substantive content which would have fit in one category (e.g. education policy), but debate records in- dicated other attributes were involved such as Negro rights. "Unique'| items included, for example, policies which apparently included a re- gional response or which were dealt with in a single roll call for the session. Of 232 roll calls, 45 were not included in the eleven prelimi- nary categories. The categories were defined more narrowly and several subdivi- sions were established. Foreign policy (A-l) items were separated from foreign trade (A-2) questions. Both suggest a liberal versus conserva- tive orientation. Foreign policy (A-I) is mostly a foreign aid dimension. It includes policy disputes dealing with technical aid and grants versus military spending; bilateral aid without “strings attached” versus con- ditional aid; aid through international agencies as opposed to bilateral aid; disarmament proposals versus proposals for military security. The content of the Foreign trade (A-2) attribute would be low tariff versus high tariff proposals; controls to benefit trading partners versus con- trols to benefit U.S. producers; encouraging payments in soft currencies rather than hard currencies. Consumer protection and Conservation implies a common anti- business bias, but can be separated into two sets of contents. Consumer 5l protection (B-I) refers to a dimension of support or opposition for measures related to pure food and drugs, standardized labeling and packaging of products, controls on loaning procedures and advertising, and protection for users public utilities. Conservation refers to pro- posals for protection of wildlife, open lands, waterways, and other natural resources as well as the enlargement of parks and recreation programs. Conservation policies can be anti-business to the extent that they prevent the private exploitation of certain resources by making them part of the public domain. I defined two kinds of agricultural policy items. C-l was viewed as a farmer versus nonfarmer dimension. It would include issues invol- ving price support levels, government assistance for cooperatives, REA, farm research and education, credit and loans, tax advantages, subsi- dized exports of farm products, etc. Protection of the ”family farm” would fit here as well. The second kind of policy content (C-2) would reflect a liberal versus conservative orientation to farm policy pro- posals, especially by those with farm interests; liberals preferring in- creased government involvement in production planning, marketing and pricing commodities, and encouragement for unionization of farm workers. I would expect conservatives to oppose controls on production and mar- keting of commodities as well as labor legislation affecting farm work- ers. Urban improvement (D) category contrasts sharply with the pro- farnuer agriculture (C-I) category. It includes proposals for federal aid to s&>lve problems which are essentially urban; such as policies for clean air, inass transportation and urban development. Instrumental proposals 52 for organizational changes such as cabinet status for agencies concerned with urban problems would also fit in this category. Negro rights (E) can be quite narrowly defined. Omnibus civil rights bills dealing with voter registration, voting, employment and accommodations for all citizens obviously fit here. So too does legis- lation regarding the Civil Rights Commission and other agencies involved in the administration and enforcement of policies Iiberalizing opportun- ities for Negroes. Civil liberties (F) is a more complex universe. It presents a confrontation of liberal and conservative values and priorities. Pro- ponents of liberal values prefer individual liberty to governmental authority, newly derived statements of values to traditional ones, indi- vidual choice to the imposition of conventional morality, secular stan- dards rather than those of the Judeo-Christian religious tradition, freedom of association rather than anti-communism. Conservative orien- tations are a basis for concern about system maintenance, whereas lib- eral values anticipate social improvements through innovation. As sys- tem maintainers, conservatives prefer an integrated and insulated society. They probably are more nationalistic than liberals. In the recent past, including the term under analysis, communism has been par- ticularly threatening to conservatives precisely because it claims to be ea revolutionary doctrine. The response of liberals to the real or imag- ined threat of communism has not been different in kind as much as it has {been different in degree. Liberals do not perceive communism to be as; threatening to the system as conservatives, thus they have not been incliiwed to anti-communistic responses with a nationalistic fervor. This I. 1"! R I ‘i ‘I 53 description suggests some correspondence to aspects of the foreign policy category. Education improvement (G) refers to policies for enlargement of educational facilities and for extension of educational opportunities at all levels. This includes grants and loans to individuals as well as developmental projects and existing educational institutions. Orientation to professionalism (H) refers to proposals which would enhance the professionalization of services for the public. Items of such a dimension may include proposals to achieve scientific manage- ment of governmental administration and services. To some extent they may reflect the principles of reformism and progressivism which came in- to vogue in municipal government after the turn of the century.h This would SUggest items concerned with the protection and enlargement of civil service, more staff experts for congressional committees, emphasis on qualifications of public employees and efforts to improve their status and pay. It might be reasonable to expect that support for profession- alism would be clearly independent of partisan identifications of the representatives than some of the other policy areas. In the tradition of the reform movement such issues may enjoy the status of being ”above” partisan politics. Part of or parallel to the promotion of profession- alism is the promotion of growth and development. Growth is equated with betterment. The entire nation must grow or die, and growth is enhanced lay professionalization (compare with Williams and Adrian, I963, pp. 2l-4I and Eassim) . “Charles R. Adrian reviews the development, assumptions and conse- quences of the reform movement (I96l, pp. 79-85 and passim). John Porter East: (I965) reviews the assumptions and value premises of the movement especially in the efforts of Richard S.(flfilcb, the major proponent of council-manager municipal government. 54 Economic policy constitutes a very broad category. There is the difficulty that most policy proposals, whether they deal with Negro rights, education, agriculture or some other subject, they also have implications for spending. Any item including a proposal to spend is subject to conflict over whether to spend more or less. But often this consideration is buried or preceded by a more dramatic conflict. The conflict concerns how the government should spend, because the spending affects two congeries of interest--labor and business. If there is a labor versus business dimension in congressional voting, certain aspects of the following characteristics can be expected. (This is not to say that all labor interest groups or business interest group as would agree on the implications or even the premises of each, but that they suggest the range of conflict.) Unions have wanted government to intercede in their behalf in conflicts with management; management desires government to restrain unions. More broadly, labor has pressed government to inter- vene in the business cycle while business has claimed to prefer an essen- tially free market. Labor wants full employment, through government spending, if necessary. Business, tending to view labor as a commodity, has preferred moderate unemployment. Labor has been tolerant of deficit spending by government and is not concerned by creeping inflation. Busi- ness has urged balanced budgets, at least over the terms of economic cycles, and wants monetary stability. Labor argues the merits of in- creased wages, including legislatedinhfinumiwage levels, to achieve econ- omic growth, but business prefers wage levels resulting from relatively freeitnarket forces. In the labor view inheritance taxes and progressive inconms taxes are means for obtaining a just and equitable distribution 55 of wealth. For business and businessmen such taxes are confiscatory and threaten the sources of needed risk capital. In general, labor is sym- pathetic to increased governmental spending and business prefers spending restraint. The social welfare (J) proposals may yield dimensional patterns of response. On one end of the dimension is social welfare and secu- rity; on the other is individual welfare and security. The value ques- tion is whether the needy members of the society are to be supported by the society collectively through the authoritative decisions of government, or whether the responsibility is a matter of choice by pri- vate groups and individuals. It is apparent that government is involved in providing welfare, so the policy questions become ones of whether or not to broaden this involvement. This may take the form of new benefits for those already defined as needy, or enlarging the definition of what constitutes a needy person. I will look for proposals affecting aid to the poor, social security, medical aid, health insurance, slum clearance, unemployment benefits, vocational training, mental retardation, etc. The final category of votes are those which are matters of party loyalty (K). This is a difficult category to characterize. The par- ties do not crassly propose and vote policies which debilitate one party to the advantage of the other. The substantive content of proposals rarely has as its object the position of the parties in the society at large. There is, of course the biennial vote for Speaker of the House. This is the only regular House decision having direct consequences on the parties and it is usually regarded as ritual observance rather than a considered response. Most studies of party voting simply use items 56 which fit an operational definition based upon voting outcome. The definition is in terms of the percentage of one party's members opposing a percentage of the other party. Lowell's (I901) criterion for a party vote was one with 90 percent of one party opposing 90 percent of the other. The Conqressional Quarterly, Jewell (I962), and others have used all votes in which a majority of each party opposes the other. I have tried to categorize in terms of the issue content of the bills and the points which generated debate in hearings and on the floor. Occasionally disputes clearly identify alignments demonstrating party affiliation rather than commitment to the content of the policy proposal. Frequently these alignments are revealed in roll calls on procedural questions which govern how or when a policy proposal will be taken up. Another cue for discovering which procedural motions get a party response is the identity of the person moving the question. Often it is the party floor leader who moves, or opposes moving, a procedural question which is delaying, embarrassing, tactically advantageous, protesting, enlightening, enter- taining, or several of the above. As Roland Young (I958, p. I60) has pointed out, there is an element of partisanship in every vote, but the significance of this element varies. I have chosen the roll calls that definitely seem to indicate, both by the substance of the debate and by the record of the vote, that the response was a partisan one. I have no new technique for identifying hitherto unrecognized party votes, Rather, I have taken apparent party divisions which show some splinter- ing by both Republicans and Democrats to get an idea about which members tend to lean in which direction when the appeals for party unity seem to be present. The party loyalty category really is a semi-residual 57 category. That is, I expected some party votes, but looked first at the substance of the issue before the debate or the vote on it. If I were to err in the placement of items, I would rather that the party loyalty category be too narrowly defined than that it be too broadly defined. Finally, the unanimous items were not useful for scaling because they allow no discrimination. They are not listed with the preliminary categories of votes. These categories are in Appendix A. The scaling process was carried out on a CDC 3600 computer using the BMD 055 program for Guttman scaling.5 This program can treat as many as twenty-five items and a large number of respondents (the number of respondents which can be treated depends upon the number items used). In the House of the 88th Congress 443 individuals were eligible to vote during the two year term. The main usefulness of the scaling program for processing roll call votes is that it records the responses of each representative on the designated items, roughly ordering the items and the respondents in terms of their support for the hypothesized dimension. This includes reflecting the responses to negatively stated motions; that is, the voted response may be “yes,” but it may constitute opposi- tion to the ”liberal” position. For example, roll call l3,l964 is among the Foreign policy items (A-l). The motion is to recommit an author- ization for the United States' contribution to the International Develop- ment Association. .The yes vote indicates a ”conservative” response to foreign,policy. 5W. J. Dixon (ed.) BMD: Biomedical Computer Programs (Los Angeles: Health Services Computing Facility, Department of Preven- tive Medicine and Public Health, School of Medicine, University of California, Los Angeles. January I, I964; Revised September I, I965.) pp. 379-389. 58 Unfortunately the BMD scaling program is not fully adequate for producing valid scale scores and correct error totals on the data being analyzed here. The problem is in the nature of legislative voting data. The BMD scaling program orders the items according to the increasing frequency of affirmative (or reflected negative) votes. Because the incidence of absenteeism varies from roll call to roll call, the items can be more appropriately ordered according to the increasing proportion of affirmative votes. The Negro rights (E) provide an example. Roll call 33, I963, received 275 votes in support of Negro rights. The scale program places this item before (to the left of) RC (Roll Call) 72, I963, with 285 votes and RC 96, I963, with 288 votes. Examination of the vote pattern reveals many apparently inconsistent votes. However, if the items are ordered by the proportion principle, a different pat- tern emerges. RC 96, I963, with the most pro-Negro rights votes also was opposed by the most votes. RC 96, I963, got 288 pro-votes out of 4l9, or 68.74 percent.6 RC 72, I963, got 285 of 375 votes, or 76.00 percent pro-votes, and RC 33, I963, got 275 of 327 or 84.IO percent pro- votes. The order of these three items had to be reversed. Because the BMD scaling program derives the scale scores and error counts from an inappropriate order, the coefficients of reproducibility calculated by the computer are also incorrect. The use of proportions indicates an assumption that was made con- cerning absences. I assumed that the proportion of yes votes determined 6The numbers and percentages reported here do not use the actual votes alone. This is explained in the text following and footnote 7. lfithher, these are the ''net” figures which contain a correction for rep- resentatives who did not participate on enough roll calls to be assigned a scale score. 59 item position regardless of which individuals failed to participate (with one exception which will be noted later). Absences were treated as occurring randomly, not systematically; it was assumed that those who did not vote would have divided in the same proportions as those who did vote. In fact, there may be systematic absenteeism. Perhaps leaders call issues to a vote when opponents are absent. If this is true the numbers of inconsistencies in my scales will be higher than they might be. Before presenting my scales, a word on the roll call data is in order. The voting record is taken from the Congressional Quarterly Weekly Reports for I963 and I964.7 I included actual votes, live pairs, and declarations made to the £9 poll. This gave me information on ab- sentees which would otherwise have been lost. Of course, it is neces- sary to assume that the declared position is consistent with the way the representative would have voted had he been present. Someone might object that this is a risky assumption and that representatives avoid painful decisions by not voting. If this is true, I do not think they would immediately reveal preferences which do go on the public record. On the other hand, I chose not to use the data recorded in the C9 An- nual Report. Here declared positions are recorded well after the roll call is past. In the interim representatives who have not voted and 7These data were supplied on punched IBM cards by the Inter-- lhwiversity Consortium for Political Research, Ann Arbor, Michigan. 'The punched code was as follows: Announced Yea Announced Nay General pair or absent Not seated in Congress at the time of roll call or Speaker of the House Yea Nay Paired Yea . Paired Nay O\lo\\n 60 have not revealed their positions at the time of the vote can assess the question with the benefit of hindsight, choosing the response which looks good on the record. I wanted to choose a prudent research strategy. I chose the Weekly Report record as providing the optimum amount of re- liable information. Torgerson's criteria for acceptable Guttman scales have been dis- cussed above. However, these criteria are largely conventional and should not be applied without some consideration of what they are for. In this research I want to investigate not only whether particular dimensions of voting behavior can be inferred and validated by scale analysis; but further, I want to use these scales for measuring in- dividual behavior. For my purposes this study is not one which is de- signed to identify all the dimensions of behavior evidenced by voting in the 88th Congress. Neither will great effort be expended in using scaling to determine whether certain, possibly ambiguous, proposals get a better fit in one scale than another. Assessing dimensionality is not, in this study, an end in itself; rather, each scale allows ranking the members of Congress on a scalable attribute. As pointed out earlier, the roll call voting of individual congressmen is what I seek to explain. I do not seek to explain the dimensions of voting. Roll call voting is the dependent variable in the constituency-representative relationship under study. The scale rankings are the data for that study. For my 'The work deck that I used was a computer recoding of the above deck. As paired for scaling. l. Yea, Paired Yea, Announced Yea 2. Nay, Paired Nay, Announced Nay 0. General pair, absent, not seated, or Speaker 6l purposes the dimensions used should be as strong as possible. The pri- mary tool for assessing that strength is the Coefficient of Reprodu- cibility, which was discussed earlier. It is clear from the derivation however, that overall reproducibility for the scale is, as Torgerson (I958, p. 3l9) notes, ”simply the average of the individual item repro- ducibilities.” To insure more strongly that the roll calls of each scale lie on a single dimension, and that the content of the scale can easily be inferred, I have decided to use a more stringent standard of item inclusion than the one suggested by Torgerson and others. No items will be included in a scale if it includes nonscale responses ex- ceeding ten percent of the total responses; that is, an item repro- ducibility of .9 or better, rather than the conventional .85. To prevent capitalizing on items with extreme marginal values to obtain a scale of CR of over .9, I will compute the CR using only the item marginals with a minority response of 20 percent or more._ I will also report Menzel's Coefficient of Scalability (I953) which is not in- flated by extreme item marginals because it uses the sum of the smaller marginal totals as the divisor for the sum of the nonscale responses. That is, if a scale item has 4l0 responses and only 30 are negative, then the 30 is used in the divisor. l 2M5 I-e>N CS=]_ l is the total of all the nonscale responses in the scale, MS is the smaller item marginal total, and N is the number of items in the scale. The appropriate level of acceptance for CS, as suggested by Menzel (I953), is ”somewhere between .60 and .65.” The criterion that error 62 patterns within a scale should be I'random" has not been clearly speci- fied. Torgerson (I958, p. 324) says: ”Practically, this means that no large number of subjects should be found who all have the same nonscale pattern of responses. If this does occur, it is evidence that more than one variable is involved.” I will not eliminate items by this cri- teria because I have raised the minimum item reproducibility to .9. By definition no more than lO percent of the respondents could then have the same nonscale pattern of response. However, I shall be alert to this kind of pattern and point it out minor manifestations if they occur within the inconsistency limits already mentioned. This scaling procedure involved submitting the selected items for computer analysis and examining the scalability of the items in terms of the criteria. Two problems are implicit in the procedure. One is deal- ing with absences. I used the votes cast by 443 members. Some died or resigned within the term. Those chosen as replacements voted on some proportion of the roll calls. Then there is the usual absenteeism by the members. Participation ranged from a low by Representative Charles A. Buckley (D NY-23), who voted on 29 percent of the roll calls, to a high of lOO percent participation by several members. Speaker McCormack (D Mass-9) voted on none'of the measures. I decided to remove from each scale those individuals who participated on less than half the items in the scale. Many of the absences have no consequence in estab- lishing a scale score. This is true as long as the absences do not in- tervene between the cutting points of the scale. An example makes this Clearer. Clarence D. Cannon (D Mo-9) died during the term of the 88th congress. But he participated on enough items to be scored on several 63 scales. 0n the Foreign policy scale (A-l) his voting pattern looked like this (7 indicates affirmative votes, I indicates negative votes, and 0 means nonparticipation): O l I l O O l 8 I 7 O O 7 O 7 O 7 There were l5 items and their order was defined by the proportion of yeas in the total responses to each item. In spite of seven nonparti- cipations, the vote pattern indicates the point on the dimension where opposition ends and support begins. Cannon's support score (8) is de- termined by counting from the right to the left. He voted affirmatively for the 8th item, and opposed the 9th item. The cutting point in Cannon's support is between these two items. It is assumed that if he were present for items 2,4,6, and 7 he would have supported them. Similarly he would have opposed items l0, II, and IS. The matter is not resolved as neatly when absences intervene on the cutting point. Repre- sentative Milliken (R Pa-7) presents a case in point, again on the Foreign policy scale (A-l): l I l I I I O l l I O l 7 7 He affirmed items one and two. He opposed item four and all the other items on which he participated. There is no evidence about how he wanted * As pointed out previously in the text, scale analysis allows ordinal level measurement, not interval level measurement. The pattern 0f zeros, ones, and sevens given above reproduces the appearance of the Fomputer printout. This pattern may leave the false impression that the Items are at equal intervals and that therefore the difference between scale scores of 5 and 3 is equal to that between scores of 8 and 6. The following is more appropriate. Assuming for a moment a di- Tension,of behavior on which the location of items (#26, 34, I4, etc.) In relation to one another is perfectly known, and the vote responses to the items by a set of representatives (A, B, C, etc.) is also known, the meaning of scale scores can be explained more clearly. 64 to vote on the third item. I arbitrarily assigned him a score of 2.5.9 The second problem is more disturbing. It involves inconsis- tencies which occur at the cutting point. Consider Representative Flynt's (D Ga-4) voting on Foreign policy (A-l): OII||777III777I77 In establishing cutting points the analyst makes the decision which re- sults in the fewest inconsistencies. For example, one would not assign Flynt a score of 2, simply because he opposed item 3. All subsequent affirmative votes would then be considered nonscale responses and there would be six of them. Clearly, the response to item 3 is an inconsis- tency. The real difficulty is whether to cut between items 6 and 7 or #14 #2I #34 #26 , , Support_ L L 3f; fa 0pp0$ltl0n + A 7 0,3 ’ C,F,G H E — E voted in support of none of the items. He voted against #26 and the others or he could have been absent on #l4 or #2l. He is assigned a scale score of O. H supported item #26 but opposed all the items to the left of #26, which included more liberal proposals. H is assigned a score of l. C, F, and G endorsed two items, #26 and #34, but not the others. Each is scored 2. Similarly D and B are 3's and A receives a Only the items and the responses to them provide markers in terms of which the respondents' ideal points can be estimated. C, F, and G get the same score, which simply indicates that their ideal re- SPOnse points are somewhere between where items #2l and #34 cut the dimension. It is theoretically likely that C, F, and G are not pre- cisely alike in the support they would give to ”all possible” proposals which might fall on this dimension. The data only warrant saying that they would support items falling into the right of item #34 and oppose those left of #Zl if they responded in a consistent fashion. H is ”more liberal” than E but how much more liberal remains un- hnown. C, F, and G are more liberal than E too. How much more liberal '5 not known, but the difference is greater than that between H and E. Scale analysis allows an imprecise but valid ordinal level of measurement. COmPare this discussion to MacRae, I958, p. 2l9 ff)- 9This procedure is arbitrary but not unreasonable. It is consis- tent with Schubert's treatment of nonparticipations by justices of the :gpreme Court. See his explanation in the Judicial Mind (I965, pp. 77, 65 between items l2 and I3. Either cut will result in four inconsisten- cies but there is a considerable difference in two scores--6 compared to l2. lntuitively it might seem reasonable to simply assign a score of 9. But in terms of scaling theory this is inappropriate. It would leave the respondent with seven inconsistencies. The choice must be made. To make it the following procedure was followed. All respon- dents who could be scored without occurrence of the nonscale response ambiguity were scored. The median scale score was determined for the group. Then the respondents who could not be scored because of ambig- uous nonscale responses were assigned the scores more (or most) like the median for unambiguous respondents. The median score for respondents unambiguously scaled on Foreign policy (A-I) was IZ. Flynt was as- signed a l2 and the responses to items 3, 7, 8, and 9 were counted as inconsistent votes. This example raises the question of how large a discrepancy between possible scale scores can justifiably be determined by this method. The literature is not clear on this point. I assigned a score if the number of scale ranks between the alternatives (l2-6=6) did not exceed half the possible scale positions (in this case, 8). This criterion is not particularly stringent.IO However, the example shown above is from a relatively large scale. Others will be discussed below which have only a few items. The rule of thumb | used maximizes the placement of respondents. loCompare to MacRae (I958, Appendix A, p. 321, 322). When ab- sences occurred at cutting points no scale score was assigned if possible scores ”spanned more than half the possible scale types.” But with ”errors” no score was assigned ”when more than one error response occurred.” However, many of his items were contrived from more than one roll call. an: u.‘ uul \L. 41. \‘\ P! (I! 66 Assigning scale scores requires two passes through the list of respondents. In the first pass those which are not ambiguous because of nonscale responses are scored. Most of the respondents can be scored in this pass and most of the non-scale responses in the overall scale pat- tern can be identified. After such a pass I total the nonscale re- sponses for each item. Although additional nonscale responses are identified in the second pass, none are removed. Therefore occasionally it is apparent after only one pass that one or more items contain non- scale responses exceeding IO percent of the entire response pattern for the item. The entire procedure of assigning scores is unnecessary. The obviously inappropriate items are removed and the remaining set can be analyzed anew. On the other hand it may not become apparent that an in- appropriate item has been included until the totaling of nonscale respon- ses after the second pass. After appropriate exclusions the set is an— alyzed again. Scaling Results The dimensions under examination and the items initially identi- fied as elements of these dimensions have already been presented. How- ever, in the steps taken toward establishing the existence of dimen- sions, some of the items were discarded and others put into different universes of items. As previously noted, it is easier to describe the central thrust of a dimension than it is to specify which items will scale together before examining the response patterns in detail. Even theri it is well to note MacRae's cautioning remark: ”[Tjhe appearance of expected [scalar] response patterns does not guarantee perfect uni- dinnywsionality. It is more reasonable to infer, if these patterns are 67 found, that we have taken an important step toward measuring a single dimension. . . .” (I958, p. 22l). Some readers may find it useful if I briefly describe how each scale was developed, the items which were dropped, and the meaning of the dimensions that are the products of the analysis. These descriptions are taken in the order in which the cate- gories were previously described. The Foreign policy scale contains l5 of the 20 items originally hypothesized. While it may seem that the elimination of one-fourth of the original items is a discouragingly high proportion, I do not think this is true for this scale. Four of the items were eliminated, not because more than IO percent of the responses on each were of a non- scale kind, but because one of the two response categories (yea/nay) of the items contained more inconsistent than consistent responses. This can occur, of course, only with items having extreme marginals; items with a percentage imbalance more extreme than ID to 90 or 90 to ID. Three of the items eliminated were those having only a single dissentor (I964: ll, 78, and 92). These single dissents were, in each case, inconsistent with the dissentors' own votes on the other items in the scale. Each had affirmed other more liberal items, but objected to one of these items. A fourth item (I963: IO) had only 32 dissentors but inore than half (l8) of the dissents were inconsistent with the rest of their item responses. The remaining item (l964: 60) which did not scale \mas very close to the item reproducibility criterion (.9). There were 4() inconsistencies out of 399 responses. The fifteen remaining items ll fowwn an acceptable scale. (See Appendix B for the specific content HThere is great similarity between this scale and Leroy N. Rieselbach's (I966) foreign aid scale. Reiselbach used eight of the 68 of the items.) The Foreign policy dimension is the basis for ranking the support in the House for increasing the United States' involvement in economic and social modernization of foreign countries. The tight- ness of the scale is evident from the coefficients of reproducibility (CR = .950) and scalability (cs = .880). The Foreign trade scale (A-2) is only a three item scale. On two of the items passage was in serious question, but on the other item the opposition amounted to less than three percent of the participants. Two-thirds of the inconsistent responses occurred on one of the three items. The scale scores for individual representatives are of dubious reliability in spite of the fact that the reproducibility indices are very high (CR = .985; C5 = .958). The consumer protection scale (B-l) is the only other three item scale treated. Four Consumer protection items were hypothesized. One (I963: 86), a proposal to establish a District of Columbia Alcoholic Beverage Control Board as an independent agency, received responses dif- ferent from the pattern established in the other three items. This might be a response to the proposal as a liquor bill, or it may be an expression of hostility toward independent agencies controlling significant aspects of urban Washington affairs. The remaining items scale tightly but two of the votes are on the same bill, one allowing an open rule for debate, the other a vote to recommit. The low proportion of nonscale responses same items; all those dealing with authorization and appropriation of foreign aid funds (I963: 6], 62, IOO, ll2 and I964: 5], 52, 6|, 62). However, he treated disarmament, as posed in a roll call supplying money to operate the Arms Control and Disarmament Agency (I963: 95), as a different issue. I find that this item scales acceptably with the aid authorization and appropriation votes in spite of the fact that its content is somewhat different from the authorization and appropriation of foreign aid funds. 69 suggests that scalability will not be a problem if in other congresses more items for this dimension come to the floor for decision. The re- producibility indices are high (CR = .993; CS = .949). Seven items were identified as conservation votes. Two (I964: 68, 73) were discarded during the scaling procedure. Both had extreme marginals one with only a single dissonant vote and the other with but eight. The remaining items revealed a well structured voting pattern. The content of these five items in the Conservation scale (B-2) are clearly concerned with conservation and the preservation of outdoor recreation space. The reproducibility of the scale is high (CR = .972; CS = .9l9). Even before reviewing the issues voted on in the House of the 88th Congress I anticipated two dimensions of Agricultural policy: one pitting farmers against nonfarmers, the other being a liberal versus con- servative Farm policy dimension. On the latter both farmers and nonfar- mers would be divided among themselves as liberals and conservatives. The evidence did not warrant this assumption. I thought I sensed the farmer versus nonfarmer cleavage on one appropriation item (I963: 4), but the remaining items seemed to be of a liberal versus conservative kind. Nine items were examined together first. Three (I963: 3l, 84; I964: 93) dealing with Mexican farm labor were obviously different from the remaining items. To the six remaining items I added the appropria- tion item (I963: 4) and constructed a single seven item scale. This scale and the individual scale scores must be interpreted with caution. It is flawed by the nature of Agricultural policy debate (compare with Jones, l96l, for example). In theory it might be expected that the 70 liberals would urge the extension of federal subsidies and price sup- ports, controls, and allotments in all farm production. However, there are crop and regional considerations which enter into the voting of the representatives. The cotton, peanuts, and rice interests are sometimes in conflict with wheat, feed grains, and livestock interests. These cleavages are evident within the agricultural policy scale (C) which I have constructed. Because these data deserve further analysis I allowed one inappropriate item to remain in the scale. A cotton subsidy author- ization bill (I963: 99) was kept in the scale in spite of the fact that the item reproducibility was below the conventional level (CR = .893 in- stead of .9 or more). The repreducibility of the scale, including the cotton subsidy item, is high enough to be accepted (CR .943; CS = .862). The Urban improvement dimension (D) is perhaps a harbinger of increasing activity and conflict in Congress. Eleven items were hypothe- sized for this dimension and two (I963: lOl; I964: l02) were removed before construction of the final scale. The first of these items was an authorization of funds for mass transit in the District of Columbia and this item clearly does not scale with the other items in the set. The second item is peripheral to the content of the scale. It estab- lished a twenty-five member national council on the arts in the execu- tive office of the President. Because the arts seem to me to be a par- ticularly urban expression, I originally placed the item in this category. The scale also includes three votes on one particular bill concerning the expansion of federal funds for public library services in urban areas. One of these items, the vote on final passage, was positioned in the scale adjacent to the arts council item. The cutting points logically 7l necessitated by this item order also brought out apparent inconsisten- cies in the library services item. Considering only the inconsistency pattern, the library vote could have been removed; however, because two amendments to the library services bill patterned appropriately and this bill seemed squarely in the dimension according to content, I de- cided that the inconsistencies were simply artifacts of an inappropriate placement of the arts council item. Removing the latter a solid nine item scale of Urban improvements resulted (CR = .958; C5 = .894). One item (I964: 64) was removed from the hypothesized universe of content for the Negro rights scale (E). This politically potent di- mension did not include many items, but the historic Civil Rights Act of I964 was twice brought to a vote in the House of Representatives. That bill originated in the House and was passed February IO, I964. It was necessary for the House to vote on it again July 2, I964, accepting Senate amendments. That timing affected the original selection of items for the scale. The non-scale item (I964: 64) was an authorization for a House committee to investigate campaign expenditures in the I964 elec- tion. Some southern representatives feared that this committee might involve itself in the Civil Rights question because the Civil Rights Act would increase Negro voting in the south. This fear was doubtless inspired partly by the fact that the House committee resolution was also voted on July 2, I964. It is clear, however, that many opponents of the House investigation committee were not opposed to Negro rights, and that many southern opponents of Negro rights favored the existance of the committee. The five item scale on Negro rights is extraordinarily tight. Very few inconsistent responses were made by the representatives on this 72 scale. Of all the scales constructed, the Negro rights scale has the highest indices of reproducibility (CR = .992; CS = .970). Thirteen of the eighteen items originally classed as Civil liberties items were included in the final Civil liberties scale and one other item was added to this group for a total of fourteen items. As I conceptualized this dimension I was looking for issues in which individual liberties, in a broad sense, were at stake. In the exami- nation of issues dealt with in the House, it appeared to me that the opposite pole of this dimension is the enhancement of collective solid- arity, or nationalism. The content of the items is superficially diverse, including items such as the House Un-American Activities authorization, a proposal to deny jurisdiction over state legislative apportionment to federal courts, and authorization for the President to proclaim Winston Churchill an honorary citizen of the United States (Appendix 8 gives the content of the items). Two of the eighteen items were eliminated on the first analysis pass because they obviously did not fit with the others. One (I963: 8) involved the issue of research and development of the RS-7O bomber and the other (I963: 39) concerned suspension of the equal time provision for all political candidates in the I964 Presi- dential campaign. After complete scale analysis it was apparent that three items (I963: 6, 29, l05) with extreme marginal totals included inconsistent votes on more than half of the responses of the smaller category. The Civil liberties dimension is not heavily freighted with is- sues raising the specter of communism. Three concern external communism. The roll calls (I963: lll; I964: 43 and l05) struck at whether 73 communist countries should be allowed to receive farm surpluses or U.S. commodities which were in any way subsidized by the federal govern- ment. Two (I963: 5, 22) tangentially dealt with the internal threat of communism: the first raised the question of whether the Secretary of De- fense ought to be able to fire security risks without appeal; the second was the vote on the HUAC authorization. For other terms of Congress is- sues related to the threat of communism might be a basis for distinct dimensions which would be more narrowly defined than the present Civil liberties dimension. One exceptional point should be noted about this scale and the order of items in it before reporting the coefficients of reproducibility and scalability. Scale items were conventionally ordered according to the proportion of votes in favor of civil liberties, enlarged foreign involvement, Negro rights, etc. As explained above, proportions were used rather than absolute numbers because of the variable of absenteeism. The number of absentees was assumed to have no effect on the proportions of representatives voting pro or con. However, in this scale such was not the case. Two items with similar proportions of pro votes (I963: III with 44.02 percent favoring the civil liberties position and I964: lOl with 44.53 percent) introduced more inconsistencies to the scale when analyzed in the usual order than they did when the order was reversed. Therefore I reversed the order, reducing inconsistencies, and assigned scale scores according to this reversal. For some reason not apparent to me, fewer opponents of the more libertarian position were absent on I963: Ill (concerning commodity credits to communist countries) than were absent on I964: lOl (proposal to deny jurisdiction over state legislative apportionment to federal courts). The reproducibility of the scale is 74 clearly acceptable (CR = .952; C5 = .850). Only six items were assigned to the education category (G). The proposals generally offer the opportunity to enlarge, extend, or improve education and educational opportunities through government assistance. The only item excluded from the hypothesized universe of content was the vote on the adoption of a conference report for a bill previously passed by the House (I963: ll2). Slightly more than half of the twenty-three votes cast against accepting the report were inconsistent with voting on other scale items. The scale was reconstructed with five items and it demonstrates acceptable reproducibility (CR = .949; CS = .8l6) although there are minor patterns in the inconsistencies which may be worth some further examination at a later time. The category of roll call votes which demonstrated the least structure was that hypothesized as Orientation to professionalism (H). Initially the category contained fifteen items, but in the process of winnowing out items containing apparent inconsistencies more items were discarded than were used. Having argued the utility of using voting scores based upon a unidimensional variable, it is appropriate to have discarded items containing sizeable proportions of inconsistent respon- ses. Nonetheless, it should be noted that scale analysis is not a tech- t al., nique for defining the universe of content. Guttman (Stouffer, I950, p. 85) has pointed this out saying: Scale analysis as such gives no judgment on content; itgpresumes that theggniverse of content is already defined, and merely test whether or not the area is representable by a single variable. It might serve as an auxiliary argument with respect to content in the special case where there is controversy over but one or two items of a large sample of items in which the remaining items are scale- able. . . . Sheer scalability is not sufficient; an item may hap- pen to scale with an area, and yet not have the content defining the area--it may be a correlate rather than part of the definition. 75 If an underlying attitude or set of systematically related atti- tudes account for cumulative patterns of voting behavior, the apparent inconsistencies in this hypothesized scale suggest the representatives' attitudes toward the role of government in promoting professionalization in governmental activities are less distinct or less salient than a good many others. The scale first constructed with fifteen items Qualified only as a quasi scale (CR = .896; CS = .689). Four items with the most nonscale responses were removed (I963: 46; I964: I9, 20, 2I), yielding a somewhat tighter scale (CR =.9I4; C5 = .732), but one in which there were some noticeable nonscale response patterns. Three more items (l964: IO, 22, 67) in which nonscale responses made up more than ten percent of the responses were removed. This eight item scale was much stronger (CR - .942; CS 3 .8I4), but two of the items with extreme marginals had half or more nonscale responses in the answer category with fewer respon- ses (I963: 70; I964: l7). Thus the scale scores of individual be- havior are of questionable reliability because of the process used in constructing the final six item scale. The reproducibility is acceptable after this process of elimination (CR = .944; CS = .825), and the con- tent of the items still evidence a concern for professionalism. The results with the hypothesized Labor versus business dimension (I-I) were also somewhat disappointing. Before reviewing the bills which came to roll call votes I anticipated that there would be several items that would sharply divide supporters of business and supporters of Labor on a single dimension. That these are polar ends to a single dimension is supported not only by conventional opinions of journalists, but by political research as well. Turner (l95l, pp. 58-64), for example, noted that the opponents on these issues were consistent opponents and that in 76 large part Republicans and Democrats were distinguishable on this dimension. The voting structure, or lack of it, in this content area is partly accounted for by substance of the items which were voted in the 88th Congress. There were few direct conflicts or ”gut” issues which set labor opposite business interests. The often threatened more to repeal section I3B of the Taft-Hartley Act would be such an issue, but this proposal did not come to the floor during the period under review. Instead there were more oblique conflicts or conflicts in which more dimensions than one were involved. Two bills on Mexican farm labor (I963: 3l, 84) previously tried in the agriculture and social welfare scales, did not fit here. A vote on the extension of federal inspec- tion and safety requirements of the Federal Coal Mine Safety Act to small operators did not scale--possibly implying that there is a dif- ferent response to proposals affecting small businesses. The hypothe- sized content, fifteen items, failed to scale acceptably. One other item (I964: 77), besides the three just mentioned, was removed. It was an item affecting the proportion of funds, spent for naval ship maintainance, which would be spent in private shipyards rather than government ones. Besides being somewhat peripheral to the hypothesized dimension, the situation of the vote may have affected the issue. The motion called for the House to recede from its disagreement with a Senate amendment. Expectably, representatives whose districts included govern- ment shipyards opposed the move. Repeating the scaling procedure with eleven items, two more did not meet the minimum item reproducibility criterion. The tenth annual one-year extension of temporary excise and corporation income taxes (I963: 37) did not fit the pattern of the other items. Several Republicans and 77 a few southern.Democrats, who favored the labor position on other items, opposed the extension of the temporary taxes. 0n the other hand a num- ber of anti-labor Republicans went along on the tax extension (which, incidently, had the support of Wilbur Mills, Chairman, and Howard Baker, second ranking Republican, of the House Ways and Means Committee). The other excluded item (I964: 6) came to a vote in a complicated floor maneuver. It involved the hypothesized dimension in that the basic pro- posal was an amendment to the Davis-Bacon Act of l93l requiring that construction wages on federal projects be at the ”prevailing rate” for an area, preventing depressed wages through the importation of cheap labor by contractors. The Secretary of Labor would be required to in- clude fringe benefits (medical payments, health plans, etc.) when de- fining the prevailing rate. The actual vote was procedural. Opponents of the amendment wanted to include a provision for judicial review of the Secretary's rate determinations. To get this change considered on the floor, the rule governing floor consideration of the bill had to be enlarged. Supporters of the original amendment voted in favor of ”order- ing the previous question on the rule governing floor consideration of the bill.” A number of Republicans, who opposed other items, supported this motion; and a number of southern Democrats who favored other pro-- labor items, appear to have favored the judicial review clause. Re-- analysis of the nine items remaining showed that on final passage of the Davis-Bacon Act amendment (l964: 7), more than half of the minority op- posing the bill was voting in a nonscale fashion. Most of these apparent nonscale responses came from southern Democrats. This item was also re- moved and the scale results were obtained with only eight items. 78 The scaleable items remaining from the hypothesized Labor versus business dimension are mostly tax items. I viewed them as Labor versus business items because proponents of tax reduction (especially on what became the Revenue Act of I964; this bill came up in the House in four scaleable roll calls--(l963: 67, 68, 69; I964: l2) argued the desire- ability of taxation changes in terms of increased employment. One pub- lic works vote was for the avowed purpose of increasing employment (I963: l2). One item broadened job protection for the transit workers of companies getting federal aid (I964: 57), another affected job re- training (I963: 36). The remaining item was a pro-business proposal which was supposed to eliminate certain excise taxes to the advantage of some business interests (I964: 55). The scale developed from these eight items shows apparent dimensionality, although there is a hint at patterned inconsistencies by some Republicans who opposed a closed rule on the Revenue Act but voted for adoption of a conference report on the Act four and a half months later. It is with some reservation, there- fore, that I will treat this scale as a Labor versus business dimension (CR a .961; cs = .888). It seems that the 88th Congress escaped, whether fortuitously or by design, the polarizing issues which might be expected to pit labor interests against those of business. It is not apparent why this should be the case. Of the fifteen items originally hypothesized, the pro-- labor side narrowly lost on only three. Of the eight scaleable items, the pro-labor position lost but once. Whether or not more distinctively pro-labor propositions might have been passed cannot be determined from these data. An imposing proportion of the House members never opposed 79 the labor side. However, bolder proposals may, I suppose, have alien- ated some marginal labor supporters. On the other hand it may be that labor has more support on the floor than it does in the sub-committees, committees, and cloakrooms. I raise these side questions for more de- tailed inquiry; the data I have do not shed further light on the point. The spending scale comes out of items originally grouped to- gether simply as economic policy items. The 55 items of the economic policy category were re-examined and broken down into smaller groups of items, one of which is the spending category. While many policy pro- posals involve to some extent the question of whether to spend more or less, that consideration seems dominant in the issues treated here. Twenty-four items were used in the first analysis run. Four were elim- inated because more than lO percent of the responses were inconsistent with the other items (I963: l3, 44; I964: 90, I08). Further examin- ation showed that each of four items with extreme marginal totals had more inconsistent than consistent responses in one of its answer cate- gories (I963: 48; I964: 45, 54, l03). Two others, upon re-examination, were taken out of the spending category to be scaled with the space spending (l-3) items. The spending scale (I-2) consists mostly of au- thorizations and appropriations for government departments and to the extent that there was controversy in committees or on the floor of the House the debate was over how much should be spent, rather than the con- tent of the programs--welfare, education, etc. The reproducibility of the scale is acceptable (CR = .94l; CS = .842). The Space spending scale (I-3) is smaller than the spending scale. Six of the larger set of economic policy items dealt particularly with 8O spending by the National Aeronautics and Space Administration spending. Opponents of NASA's spending seem to have objected more to its objec- tives than its spending. These opponents sniped particularly at the objective of having a man on the moon by I970 and preferred that more emphasis be given to developing military sophistication in space re- search. The six hypothesized items scale acceptably (CR - .983; CS = .946). Substantial voting structure is evident concerning the orien- tation of the 88th Congress toward the national debt (l-4). Four times the issue came to the floor, accompanied by recommittal motions three times, and twice there were votes on consideration under a closed rule. In all there were nine items. No debt extension proposals failed, but most of the votes were close. One vote on a closed rule (I963: 56) identifies the most extreme opponents to increased national debt. The reproducibility of the scale is quite high (CR = .979; cs = .952). The Social welfare scale (J) consists of fifteen items. Seven- teen items were first hypothesized for the scale. One (I964: 69) was dropped on the first pass because of a large number of inconsistencies. Three others (I963: 3], 84; I964: 93) which deal with migrant farm labor, but which do not scale with Agriculture policy, were added for analysis. Two (I963: 3l, 84) were dropped after the first pass because of nonscale responses. After the second pass two items (I963: 55; I964: 93) which have extreme marginals were dropped because less than half the responses in the smaller categories are consistent. The re- maining items deal with welfare measures including mental retardation, vocational work study programs, increased social security benefits, and 8l housing. It might be noted that about one-third of the nonscale respon- ses in the scale are on two items. One (I963: 50) is substantially mar- ginal to the universe of content. The vote was actually procedural, but it was a rule to waive points of order against a rider to a foreign ser- vice building authorization allowing small claims against the United States by Filipinos for war damages. My decision to include it in the social welfare universe hinged on the consideration that small claims were being encouraged and settlements were to be made with individuals. The other was a fairly closely contested administration bill (I964: 30, food stamp bill authorizing $400 million over four years) on which there were seventeen nonscale yes votes. Fifteen of these seventeen were cast by Democrats. That is, their general pattern of voting would lead one to expect that on this item these fifteen Democrats would oppose the bill. These nonscale responses may be a clue to party or pressure group efforts. The reproducibility of the whole scale is acceptable (CR = .956; C5 - .873). Lastly, party loyalty dimension consists of all fourteen of the items hypothesized for the scale. Of the items six are decisions on substantive proposals, six are procedural votes moved by one party to delay, obstruct, or embarrass the opposition, and two are organizational votes (for details, see Appendix B). Further examination of the voting pattern will be made in a later chapter. For the present it should be pointed out that support for the items range from a higher proportion on the right to a lower proportion on the left. This makes the scale a dimension of support for the Democratic party--those with the lowest scores are the most loyal Republicans, those with the highest scores are 82 the most loyal Democrats, and those with intermediate scores are switch- ers. As is evident from Table lI-l below, this scale shows the most polarization of voting; 65 percent of the representatives are loyal to their parties on all of the votes constituting this scale. The firm structure of the scale is evidenced by the reproducibility indices (CR = .98l; CS 8 .953). Hypothesized Relationships between Constituency Characteristics and Scale Scores The results of the scale analysis are useful in several ways but the first use I have for them is to see whether they are related to con- stituency characteristics. The scales provide an ordinal measure of the voting behavior of the representatives on several dimensions. The general hypothesis of this study is that representatives take their con- stituencies into account in their roll call voting and that particular constituency aspects are salient to different policy dimensions. Some of these salient aspects may be indicated or approximated by available demographic and socio-economic variables. Therefore, although the causal links may be tenuous, certain associations should be apparent between the constituency variables and the scale scores of roll call voting. The analysis technique I will use is correlation, which is simply a coefficient of relationship. Direction of cause cannot be inferred from this coefficient, of course. I will hypothesize the direc- tion from the theory already presented, inferring support for the hypothe- ses if the associations I find are strong. Before hypothesizing and assessing the strength of the relation- ships between the independent and major dependent variables, some 83 m.a_ m.m: o._ m.:: N m A~io__oa _oese_so_co< m.m o._m N.m w._: o m A_i_v mmoc_m:n .m> Loam; a.m- ~.mm o.m~ «.mm m. a. Age >e_o>o_ scene ~.m m.aa N. o.ma N m. Ase ocoe_oz _o_oom N.m m.om o.m_ m.mm m_ m. A_iu__oa cm_ocou _.m awn. _.m m... N m ANimv co_um>comcou m.m «.ma m.o~ m.- N m Ace ocoao>ocaa_ cope: s.m m.~ k. _.N a a. A~-_v mc_ocoom :.m :.m m.N w.N m :— Auv mo_ucon__ __>_u >Lomoumu >L0moumu ucoocoa _+xN\oo_ o_mum o_mum on cam om Axe oLOUm u;m_c pew. coozuon mo_o 0—mom o_mUm .oeocuxo oEocuxo iccaoocc _m ecu c_ can a can u ecu c_ on» c_ ic_mLmE ;u_3 mEou_ mama co c_ m.x. mama mamm mEmu_ 0—mom $0 Ne .33 co N .eo we. mo x e0 3952 con—=32 u m a u m < mco_mcoe_v 0—mom motoum 0—m0m mc_uo> __mu __oL cooux_m ecu mo A T mo_ee_coeoococo co_ose_cen_o oco >e___eo_oon oep--.-__ meme; .moum_a _mE_uoo Lao» cu mc_um_:u_mo >n uoc_Ecouoc coccom ~m.m awn No.mo mmm. mam. A~i:98. scene mm._ am: om.mm new. mmm. Ase ocoe_oz _t_oom mm._ ems om.m~ 0mm. 0mm. A_-o__oe cm_ocoe mm._ omm em.c~ m_m. NNm. A~-mv co_oo>concoo am._ m_a ma.k~ amm. mmm. Ace ecoao>ocaa_ coat: m... Mn: we.m~ New. _am. A~-_V mc_ocoom o... mN: NN.N~ omm. Nmm. Adv no_ococ__ __>_u mo_comoumu 2\\muewv o_mum ecu mace sz c_ meme mo co_u o_mum mo co_u -se_cen_a coco e_ -se_eee_o >e___e_oso come ozu cocoon ozu co Ame >u___nm_mom iOLme mo LoLco mama mo co_um_>oc mo uco_u mo aco_u oeoocoem coeasz ocoocoem -_eeooc -_eeoou x 1 _ z o mco_mcos_e «_mum mmmaqmmmm mucoum o_mum mc_uo> __mu __OE cooux_m ecu mo mo_um_couumcm£u co_u:n_cum_o ccm >u___nm_MUm ochii_i__ m4m<fi 85 remarks will be directed to the structure in the independent variables themselves. Appendix C presents a table the simple product-moment cor- relations between each pair of constituency variables. The summary of relationships, which specifies the correlations of f .65 or more, sug- gests quite clearly the structure that is present in the constituency data. Some of these associations were anticipated in Chapter I, but some are in marked contrast to the expected values. There are two related but distinct sets of four variables each. These are the economic and educa- tion sets. Besides the expected strong associations between Median in- come (ll), Median home value (l7), and Median rent (l8) it is evident that the percent of sound housing units with all plumbing facilities (l4) measures pretty much the same underlying economic variable. The education variables, the percent of persons with less than five years of school (7). the percent of persons with four years of high school or more (8), the percent of persons with four years of college or more (9), and the median years of school completed (IO), all substantially corre- late with one another with the single exception that the percent with low education (7) does not show a Strong negative correlation with the college variable (9). Expectably, there are firm associations between these sets of items, but generally at a lower strength than the associa- tions within the sets. This supports what should have been intuitively clear already; that when the relationships between the independent vari- ables of either set and the dependent variables of roll call voting be- havior are examined, economic variables should be controlled on educa- tion, and educational variables should be controlled on economic ones. An employment variable which seems to mediate between these two sets is 86 percent of White collar workers (l9). It is most highly associated with college education (9), but it is strongly associated with all the econo- mic items and the percent urban of the total population (2) as well. The latter, percent urban (2), which I will refer to as a ”demographic” variable along with percent of population change (I), percent Negro of the total population (3), percent Foreign stock of the total population (4), and Median age (5). is substantially associated with all the econo- mic variables and is negatively related to the percent of Farm workers (2l) in addition to its relation to percent White collar. There is some mildly surprising information in these correla- tions. For example, the percent of housing units owner-occupied (l3) does not correlate very strongly, compared to the other items, with economic, education, demographic, or employment variables. It corre- lates above the .65 level only with Median rooms per housing unit (l5) and these two items form a mutual pair (i.e., of the 2l items examined, the highest correlate of each is the other). This contrasts with Mac-, Rae's (I952) interpretation of the variable in Massachusetts using I950 census data. He interpreted it as a rural-urban variable. More recently, Froman (I963, p. 9l ff.) treated it as a measure of socio—economic status. The finding here is consistent with that of Frost (I959), whose rank cor- relations of Home ownership with Urbanism, Median family income, percent of population with four or more years of college, and percent of popula- tion with professional and technical occupations shown no strong rela- tionships (the highest is .52 with percent urban of the population) and with those of Hadden and Borgatta (I965), who examined I960 census data for all (644) cities in the United States with populations over 25,000. 87 Compared with 65 variables, using the ”all cities” data, the percent of owner occfipied units correlated above the t .65 level only with ”percent housing units in one unit structures.“ (Rooms per unit was not included among their variables.) Hadden and Borgatta factor analyzed their data, producing l3 factors which they felt they could label. The Owner oc- cupancy item appears in five of these factors but its loading never ex- ceeds .45 using the ”all cities“ data. That highest loading is on the ”population density” factor. It is fairly clear then that the Owner occupancy variable is not a very clear indicator of any of the better understood sociological variables such as SES, Education, Urbanism, etc. It is interesting to note that percent Blue collar (20) is not highly correlated with any of the other variables. The variables most strongly associated with it are education variables. It correlates negatively with College education (9, with r = -.55) and High school education (i, with r = -.52). The percent of the total Labor force un- employed (l2) is not strongly related with other variables either. Its highest correlate is percent Blue collar (20, with r = .43). The high- est correlate of the percentage of Negroes (3) is .6l and that is with the low Education variable (7). The noticeable characteristic of the correlations between percent Negro and the other variables is that the signs are predominatly negative. On the other hand, percent Foreign stock (4) correlates with two economic variables at the .65 level or higher. But its highest correlate, interestingly, is the percent of persons enrolled in Private elementary schools (6, with r = .7l). This is a mutual pair of variables. I think this tends to support the sug- gestion made in Chapter I that the Private school enrollment variable 88 is roughly indicating strong religious affiliation from both Roman Catholic and Protestant supporters, many of whom are Foreign stock. Perhaps it is less surprising that population increase (I) does not highly correlate with any of the other variables. The percent of Farm labor (2|) correlates negatively with Urbanism (2, with r = -.75), and the economic variables of Median income (II, with r = -.68) and sound Housing with all plumbing (l4, with r = -.66). Median age (5) correlates strongly with the median number of persons per housing unit (16, with r = -.72). It is useful to know the intercorrelations of these variables so that the analysis of the dependent variables may be done with some parsimony. The economic variables form a tight cluster, each associated with the others of the cluster more strongly than with any of the other seventeen variables. Home value Q, (l7) 9]‘§\;S:k Homes with Median income sound plumbing s\ (ll) (l4) . 6% «2 \Median rent (I8) \ o7 O) Because this is true and because these economic variables are strongly related with other variables it will be desirable to control for econo- mic variables while examining relations between other independent vari- ables and roll call voting behavior. These controls will be easier to introduce if they are carried out with a single variable than with each of the four taken one at a time. Therefore, to select the variable 89 which best represents the underlying economic dimension I have calcu- lated mean correlation for each variable on the other three variables. The variable with the highest mean is the best estimate of the under- lying dimension. Table I|-2--Intercorrelations of the economic variables Homes with Median income sound plumbing Home value Median rent (ll) (14) (I7) (18) (11) R .90888 .85438 .90400 (I4) .90888 a .78108 .84319 (17) .85h38 .78108 a .82956 (18) .90400 .84319 .82956 * 2.66726/3 2.53315/3 2.56502/3 2 5675/3 Means .88909 .84438 .82167 .85892 Median income is the best estimate; therefore, when controlling for econ- omic effects on relationships, I will simply control on median income. A similar procedure with the Education variables is inapproriate. While the percent of the population with High school education (8) and Median education (I) are the most highly correlated mutual pair of vari- ables among all the variables examined, Low education (7) and College education (9) correlate more highly with other variables than they do with variables of the education category. This means that no single variable of the category characterizes the information well enough to serve as a control for the category. Wherever it seems useful to con- trol for education it will be necessary to control individually on each of the variables. 90 The measurement of the dependent variables-~scales of roll call voting--should also be inspected before hypothesizing relationships. To this point I have not tried to say which scales yield the best measures of voting in the respective dimensions. Only the reproducibilities have been noted. The literature is not particularly helpful here. Guttman (Stouffer, gt_gl., I950, p. 79) has SUggested that with dichotomous items ”It is probably desirable that at least ten items be used, with perhaps a lesser number being satisfactory if the marginal frequencies of several items are in the range of 30 percent to 70 percent.” This consideration bears on the scalability of the dimension in question. However, it leaves unanswered the question about which of the scales with less than ten items ought to be considered seriously. Eleven of the scales have fewer than IO items; of the eleven, fewer than half the items are in the 30--7O percent range on only the Conservation scale (B-2) and the Aid to education scale (G). It is worth mentioning, however, that in the scales presented here an unusually large number of respondents have been scaled. Nearly 6600 scale scores were specified, averaging well over 400 per scale. With the lowest CR = .94I (Spending, I-2) and the lowest CS 8 .8l6 (Aid to education, G) the scalability of the dimensions is re- markably well demonstrated. Distinguishing among the scales can probably be done better with an assessment of their discrimination than by simply considering the number of items in each. An example illustrates the point. An ideal scale for scoring 400 respondents might include 9 items. Although it is not possible to know by scaling alone the distance between the items, it might look like this: Items 9 8 7 6 5 1+ 3 '2' l "espmdents no 1.0 40 I” L0 L10 1.0 lhol 1+0 in category More important is the question, how many respondents fall into the scale categories. There are ten categories (0, l, . . . , 9). If the item dis- tribution and the respondent distribution were ideal, there would be 40 respondents, or ID percent of the total, in each scale category. Unfor- tunately the distributions in the actual scales do not resemble this ideal very closely. What'is immediately apparent about the distributions is that they tend to be restricted at one end of the dimension or the other. Usually, it is at the left end. As I have set up the scales this. means that a large proportion of the representatives supported the most extreme item on the scale and that there is no way of distinguishing which respondents might have supported an even more extreme item. Table ll-l shows the degree of curtailment for each scale. The degree of curtailment is disturbing particularly because it can be expected to produce depressed correlation coefficients (McNemar, I962, p. l44-5). Certain corrections can be applied to improve cor- relations based upon parametric measurement, but none have been developed for non-parametric ones. Table II-l also reports ideal category percents (F) and standard deviations for the observed scale score distributions (I). The repre- sentatives were ideally distributed across the scale categories of the Foreign policy (A-l) scale, for example, there would be 3.4% in each of the 3l categories and the standard deviation (S) would be 0. However, as columhs C, D, and E indicate, certain categories include far more than the ideal share. This accounts for much of the variation indicated 92 for each scale in column I. There is a further explanation for this great variation. It will be remembered that where absences intervene on the cutting points of the scales half scores are used. In general only a small proportion of the representatives received half scores. There are, therefore, more whole than half scores. Yet the existence of these half scores necessitates treating the data as if the number of scale categories is 2 K + I, with K equal to the number of items in the scale. I have reported the standard deviation simply because it pro- vides the reader with some feel of how distributions of scale scores in each scale compares with that of the others. It is obvious that where variation is high, most of the scale scores are falling into only a few of the scale categories. Comparing the Orientation to debt (I-4) scale to the Urban improvement (D) scale, the standard deviations indi- cate that the distribution is more uneven in the former than in the latter. The Consumer protection (B-l) scale with only three items shows much less variance than the six item Space spending (I-3) and Orienta- tion to professionalism (H) scales. The final coefficient reported (column K) is the standard error of the mean (Sm =,7%—', with N = the number of respondents; McNemar, I962, pp. 73-79). Conventionally, this statistic is used to make inferences about a population mean from the mean and distribution of a sample, similar to standard deviations which ordinarily are used to explain population distributions from samples. Here the purpose is simply to point out the degree of regularity in the distribution and the stability of the mean. Each of the relevant charac- teristics of the scale score distributions has weight in this statistic: the unevenness of the distribution, the number of items and the number 93 of respondents. It is noticeable, for example, that most of the scales with a small number of items (and usually a smaller number of respon- dents) have a high standard error--notice Foreign trade (A-l) Aid to education (G), Orientation to professiOnalism (H), etc. However, the Orientation to debt (I-h) scale has nine items and still has a substan- tial standard error term. On the other hand the Conservation (B-Z) scale, with five items, has one of the lowest standard error terms of the sixteen reported. These measures give a much fuller picture of the scales than merely considering the number of items and how many of the items have marginal frequencies between 30 and 70 percent. At this point it seems inappropriate to make a decision about which scales ought not be given full consideration because of the pro- portion of respondents in one or both of the extreme scale categories, the size of the standard deviation or the size of the standard error. Compared to one another, however, it is apparent that some have much better distribution characteristics and will probably yield larger coefficients of correlation and shared variance than can be expected of the others. The hypothesized relationships between the independent variables and the scale scores of congressional voting should be discussed briefly at least. In general I am hypothesizing relationships of scales with variables because I think that these variables are indicators of prefer- ences, expectations and/or interests which representatives will take in- to account when they make their roll call voting decisions. The evi- dence of relationship will be rank order correlation coefficients. I will hypothesize strong, moderate, and weak relationships--the latter 9h indicating I expect no meaningful relationship. Tentatively, I will de- fine strong to mean rs = t .hOO or more; moderate, rs = t .250 to t .399; Igggh, rs ‘ + .299 to - .249. Statistical significance is not particu- larly helpful with these correlations because the number of observations being correlated is very high. In later parts of the study signifi- cance levels may become relevant as correlations within subgroups of the House membership are examined. However, with 400 observations cor- relations much smaller than t .250 are significant beyond the .00] level. I expect that support for Civil liberties (F) will be strongly associated with indicator of substantial education (8, 9, l0) and with the percent Urban (2). The relationship will be negative with low education (7) as well as with Median age (5), Blue collar (2) and Far- mers-(2|). The relationship will be moderately positive with White collar (l9), the economic variables (ll, IA, l7, l8), Population change (I), and Foreign stock (4). I expect a moderate negative relationship with Private school enrollment (6). Weak relationships will occur with Unemployment (l2), Owner occupancy (13), Rooms per unit (15) and Persons per occupied unit (I6). I also expect the relationship with the percent of Negroes (3) will be weak, but this might not be the case if data on the south and southern representatives were excluded. The expected relationships for all the scales need not be de- scribed here. Table Il-3 summarizes them. I have hedged somewhat on predicting relationships of the four scales which yielded the least dis- crimination. The only variables mentioned are those expected to be at least moderately related to the scales. TABLE ll-3--Hypothesized relationships between 2l independent constit- uency variables and I6 roll call voting variablesa Dependent variables scale scores on Correlations hypothesized between the independent variables (l-2l) and the scale scores of roll call voting Strong Moderate Weak. Civil liberties 29.5,.2: 8, l, 4, 6, ll, 3, l2, l3, (F) 9, 10, 20, 14, 17, 18, 15, 16 21 19 Spending 2, 4, _l, l, 3, 5, 6, 7, 8, l0, l5, (1-2) 1151,11 9, 12, 11, 16 _l 19. 20 Urban improvement 2, 4, 9, 1, 3..§. 7. 3. 6. l3, '5. (o) 19, 21 10,11, 12, 16 15. 11. 1.5. 20 Conservation l, 4, 8, 9, 5,‘11, I3, 14, 2, 3, 6, (3'2) 10» Z-l 16. 11. L8... 7. 12. l5. I9 20 Foreign policy ‘5, 8, 9, 2, 4, Q, 1. I. 3. l3» '(Afl) 10, 19, 11,11, 1L1, 15, 1 , _2_1 1 , 18 20 Social welfare 2, 3, 4, 5, l, 6, 113 l , l5,,l6 (J) 7’ 8’ 9, In, ll, '8’ 1o, 12, 21 19, 20 Party loyalty 2, 3, 4, '1,.§,'§, 7, 8, l0, IS, (K) 11; '2. 15; .2,.ll l5. 2' L2. l_8. l . 20 Labor vs. business 7, 11, l2, 2, 4, 5, 6, l, 3, 8, (1-1) fl, 1 ,1§, 9, 11, 19 1o, 15, 16 20,._1 ‘ Consumer protection 2, 4, 7, 21 3, 5, 8, 9, l, 6, l5, (13-1) 10,11, 12, 16, 19 I3, _l_L‘.’ _l_Z’ 1g, 20 96 TABLE ll-3--Hypothesized relationships between 21 independent constit- uency variables and 16 roll call voting variablesa. 4* Con tinued Dependent variables scale scores on Correlations hypothesized between the independent variables (1:21) and the scale scores of roll call .voting Strong 1Moderate Weak Negro rights .5, 21 .l. 2..fl».§s.1a 3, 15, l5. (s) 8, 9, 10, 11, 19, 20. 12, 11, 14, I7, 18 Agricultural policy '_1,‘1_,.11, .1, 3, 4,'§, 8, 2, 5, 7, (c) __8_: 2] 9: lot _L. '29 '39 15: 19, 20 Orientation to debt .5, 6, Z, 9, .1, 2, 4, 8, 10, 3, 15 (I'll) 12! __l L1,: '39 B: '6’ 1 , 1_, 1 , 20 MODERATE Space spending 1, 2,‘§, 9, 21 .(1-3) 1 Orientation to pro- 1, 2,.5, 9, ll, 18, 19,.21 fessionalism (H) Aid to education 2, 8, 9, 10, l4, l7, l8, 19 (c) [Foreign trade (A-z) 5. 9. 10.1.2. .1 A aThe italicized variable numbers indicate that negative co- efficients are expected. Because allow correlation indicates no relationship, and sign becomes nearly meaningless, none of the variables hypothesized as weakly related is italicized. The hypothesized correlations are not very_high, particularly when compared with the values noted among the independent variables. But as noted before, the decision-making which underlies the depen- dent variables is very complex. The level of measurement used on 97 these dependent variables is only ordinal, thus the correlation tech- nique must be rank correlation. The correlations will be attenuated by the fact that many of the respondents have tied ranks. Finally, the distributions among the scale score ranks are often uneven, a matter which will be explained more fully in a later chapter. These measurement problems make it even more difficult to predict how strong associations will be between the independent variables and the depen- dent variables measured in the minor scales. Introducing controls to these relationships should make some clarification possible. Education, particularly College education (9) probably intervenes upon the relationship of economic variables with the policy scales. Controlled for percent of college educated, the higher the median income the more negative the relationships will be between it and the roll call voting scales. In some policy areas I expect economic variables and education variables to correlate with scale scores with opposite signs. To use the spending scale as an ex- ample, I hypothesized that income will correlate negatively and College education will correlate positively. If this is true, then in spite of the fact that the correlation coefficients may be quite close to zero, the application of statistical controls will sharply affect the coef- ficient. That is, because the Pearson product moment correlation be- tween Median income and College education correlate is .66, partialling out the effect of college education on the relationship between income and the Spending scale will dramatically increase the coefficient of cor- relation between income and Spending. It is difficult to predict how the percent of Negroes in the population (3) is related to the scales of voting. That percentage 98 has different if not opposite meaning in the south than it does in the urban areas of the north. In several scale relationships I have hypothe- sized weak relationships because I expect that the regional meanings can- cel one another out. Controlling for region will sharpen the understand- ing of how this variable is related to scales of voting. CHAPTER III THE RELATIONSHIPS BETWEEN CONSTITUENCY CHARACTERISTICS AND ROLL CALL VOTING Recently, it has become fashionable to study public policy out- comes comparatively and account for the differences in outcome by dif- ferences in economic development. These comparisons have been made at different levels; for example, Cutright (1965) has compared the social security policies of 76 nations, Dawson and Robinson (1963 and 1965) examined welfare policies in the American states, and Williams, Her- man, Liebman and Dye (1965) have studied urban services of small munic- ipalities in the Greater Philadelphia area. In these and similar studies the authors have interpreted their findings to mean that ”econ- omic development” variables explain the level of services provided through public policies better than do political variables. In general, the conclusions have been that the more developed a system is econom- ically, the higher the level of services provided through the public policies of the system. The central question of this study, whether constituency charac- teristics explain roll call voting, seems to be similar to that posed in the studies cited above. However, there are distinct differences that ought to be noted. The units of analysis are quite different. Dawson and Robinson, for example, are dealing with states and states are sys- tems unto themselves, as Herbert Jacob (Jacob and Vines, 1965, p. 8ff.) - 99 100 has pointed out. This implies a number of characteristics (see, for example, Easton, 1965a and 1965b; Mitchell, 1962) which are not associated with congressional districts or their representatives. A congressional district provides very little sense of community to those who are legal members of it. It even lacks a name. Ordinarily, it is the unit of organization for only the election of a representative and the campaign organizations with that purpose rarely number more than four at one time. At the time of the primary elections, for example, there may be two contesting nuclear organizations in each party (Schlesinger, 1966, pp. 125-127). If any of these nuclei is to persist over a very long time, it will probably be the winning one, and even that organization is seldom substantial. In a state system, the author- ities make binding decisions which materially affect its members. These authorities are responsible to those over whom they have authority. The representative does not clearly resemble state government. State au- thorities are several, while the representative is only one man. More important the representative does not make binding decisions upon his constituents as state authorities do. The representative has a voice in the decision-making which binds members to a much larger community, his constituents included. He is responsible to a small part of the larger community as he shares in the authority over the larger community. To explain national policy outcomes in the manner of the research cited, the focus would be on national resources. This is not the procedure used here. It is unlikely that measures of economic development in constit- uencies will have the explanatory power for congressional voting that they seem to have in explaining system outcomes. The resources of the lOl district are not redistributed within the district by the decisions of the representative; rather, he shares in deciding what part of the system resources will be used, and what segments of the society will be benefited in the distribution of these resources. To the extent that his constituents are differentially affected by these decisions, his voice in the decision may speak for the attitudes and interests of the differentially affected members of his congressional district. The importance of his voice in the decision which results may be of little consequence as the representative relates to his constituents. What is probably more important to him is that his voice is heard by constit- uents than that it be heeded in decision-making. Roll call voting is one mode of expression by which the congressman speaks for his district. . The vote, or even an index of voting, is not very comparable to a policy output ranking. It may more closely be compared to an individual raw score that a researcher might obtain as he seeks to determine the aver- age IQ for a large high school. The roll call vote is only one step, albeit an important one, toward the implementation of a policy output. From the findings of Dawson and Robinson and others, it seems that the availability of system resources sets important boundaries-- upward and downward limits--within which the level of outputs will fit. What is not clear from their findings is which decision-makers take into account these boundaries or whether the boundaries simply impose them- selves like some invisible hand as the decision-making process is car- ried on and the various combatants seek to work their will upon the decisions. I do not expect to find relationships between constituency resources and the voting of congressmen which are as strong as those discovered between system resources and policy outputs. Constituency 102 resources are less consequential in the sense of presenting limits to a representative. Perhaps he looks at the differential effects of policies and how these relate to constituency interests and attitudes. To the extent that these attitudes and interests are associated with constituency characteristics I will be able to detect statistical associations between constituency data and roll call voting. A second area where findings may be different from Dawson and Robinson concerns the importance of political variables. Dawson and Robinson, along with Thomas Dye (1966, pp. 238-259 and passim), have found that differences in policy outputs are not as related to party control, competitiveness, voting participation, apportionment, and other political variables of the system as they are to socio-economic variables. I have designed my research to examine relationships of constituency data with roll call voting with and without separate con- sideration of the two parties. The effects of party, competition, seniority, ambitions and other political variables will be considered, too. The variation by party will be taken up more thoroughly in the next chapter. Correlations between Independent and Dependent Variables In a statistical sense, the analysis presented here is an ele- mentary one. The correlation coefficient describes the extent to which a pair of variables vary with one another. As I have explained above and in chapters one and two, I did not expect the coefficients between the constituency characteristics and the voting scores to be very large. By Flinn's (l964) rule of thumb, a coefficient of .4 or higher should 103 be taken as a substantial relationship between an independent variable and a measuring of voting. This is to say that explaining roughly 16 percent of the variance in voting is a substantial chore for the legis- lative analyst. The relationships between the constituency variables and the roll call voting variables are presented in Table III-I. The matrix presents Spearman rho rank correlations. For Kendall's tau correlations, see Appendix 0. Of course tau, by mathematical necessity, is a lower coefficient than rho. The two measures are more similar at zero and near zero correlations, but at higher levels tau is only two-thirds the magnitude of rho.(see McNemar, 1962, pp. 203-205; Siegel, 1956, pp. 202-229). However, an advantage with tau, which will be capitalized upon later, is that it can be generalized to a partial correlation coefficient. These rank correlations are an appropriate alternative to the Pa rametric measure, Pearson product-moment, which requires equal inter- va 1 measurement and scores from a bivariate normal population. It is, however, difficult to compare the rank correlations to the product mo- men 1: ones. It is known that the "efficiency" of the rank correlation is about 91 percent. This means that to reject the hypothesis of no re 1 a t ionship (the assumptions for proper use of the product moment cor- re I a t ion having been met) between A and B, with a hundred cases rank C—or— re lation will reveal the correlation at the same significance level as t he product moment will with 91 cases. However, there is no known a ‘ Set) raic function relating rank correlations to product-moment cor- T‘e. - . . . ‘ a t IOI‘IS. For my own information, I have compared the rho correlations 104 co: Nna. :mo mud mNN _mo «Na. AN-6:8 _ntse_su_tm< mmm ONO N__ ___ MMN ONN mam. A_-_v ense_nse .n> Loans 0.: N66 mmo mmo NNN mMN mmwn Axe >:96. seen; oNN N_N Na. meN mm. mmN mm_ Ase senc_ez _s_uom om. ONN :MN :em mN_ 0.: Add. A_-6:8 em_otoc N_N mN_ mo. m_N _mo NmN «Na. AN1mv eo_un>tsneou :mN MNN OON w_m NM. mmm mmfl Aav semen>otee_ eneta mN: mNo moo NNo mmN _m_ mww AN-_V me_6eeem N6_ mNN NON NNm :N_ can we. Auv ns_pese__ __>_u co_umo:vo cozy >Lmu imuaoo icoEo_o omcmco 30— uum>_ca om< xUOum co_um_ eu_z x ;u_z N en_eoz cm_utou x otmoz x cent: x -seoe x N o m a m N _ mo_nm_cm> __mo __OL m. can emu—nm_cm> >oco:u_umcou ucuvcuauvc_ _N cuozuon mmco_um_oggoo ocz _.___ ~4m6:2. .n.s..=e..m< Nmm. NAM. m_m MNm. WNW. awq. HNN A—i_v mmoc_m:n .m> Loam; mum. aw“. oNN «Mm NeN MNH mmN Ag. >:98. >etn. mmo mam NmN soc Nam. mum «ma An. pene.oz .n.uom mON mam m.N Na. Nam soc mam. A.-6:2. em.e.ou one am 3N mg g «mm mm... 3-3 co.um>.mmcou mm a 2m 2. w.N|_ mum mm... 2: 22.98%. :35 owo wNw _oN wad mam mad mum AN-_V me.eeoem 6N. www. NeN MN. mam mmm NNm Au. np.u.oe.. ..>.u mc_ co_u ness—a mu_c: co_u imosuu __m no.3 mc___036 co_u imusoo _oosum mu_c: uo_a:uoo uo>o_a osooc_ imuauo omo__ou cm_g pesom x Lpeso a. -eoea x en.eoz ep.uoz, 56.: a. 56.2 x e. m. N. .. o. m m ..nu :6L 6. see a mu_nm_cm> >uco:u_umcou ucoocoauoc_ _N coozuun mmcobum_oggoo on: o {1 voucwucoUiimo_nm_cm> ~1_~. udm >ocu:u_umcou mo mco_ua_L0mov om_uoca ucoe LOm _i_ o_nmh moms o>_ummoc 0cm muco_u_wmmou co_um_oc.oo ouN_u__ma_m Q N. N3 NM a ..1:Io an 3-5 6.6.: e326“. mam NN. mm. mm. NN. mad. “MN. Au. co.Nnu=us op u.< an mm. N... 3.. N: Na. El. .5 3:68; imowoca ou co_umuco_co «mm JMN emo NNN. :oo Nmm wmw Am-.. me.eesen «seem NMM NON mam. wmfl mwm mmm. mmM A:i_v anon cu co_umuco_co mam moo men NN: me: mad. mNN Am. nuem.. otmsz AWN m:_ _w_ mm. mmN mwm. wMM. A_imv co_uoouoca Lusamcou mwm mm_ AMM. «mm. mmm. mmm. wflm Auv >o__oa _mcau_:u_cm< “MM MMN “Md. Nam mwm wmq mmw A_i_v mmo:_msn .m> Lona; MNM mm. mam. mum. Mam. mmmn WWW Axv >u_m>o_ >ucmm H .m. mmo mNo mo. £3 E A3 88.63 .268 pm” me. NN. 6:. m.N wad. am. ...<. >6:2. em.e.o. CNN :__ moo 0.0 mmo mwfl mum aNimv co_um>comcou WWW. m_N _o_ mmo nm_ wmm mMM. Aav ucoEo>o.aE_ cone: QON NON wwm mwfl wwm MMMH WWW ANi_V mc_ocuam “MN. mm_ md_ _m_ QON wN_ N_~ Aug mu_ucon__ __>_u “_c: u_c: me.. me: ..636 ._626 mcoxcoz muuxcoz o:_m> Lea Log mcoecmu cm__oo cm__oo ucoc ego; mcomcoa meooc N os.m N 66.53 N en.eoz en.uoz en.eox em.po: _N cu m_ w_ n_ 0— m— uo:c_HCOUiimo_nm_cm> __mu __oL m. can mu_nm_cm> >uco:u_umcou acoucoauoc_ _N couzumn omco_um~wLLou egg on a _n___ mgm<fi 107 of the constituency variables with one another to the product-moment correlations (the latter are in Appendix C). The variance between pairs of correlations I have observed tends to be between .020 and .110. Signs are different only where the relationships are very slight. There seems to be no pattern suggesting that one or the other type of correlation will be consistently higher than the other. Table 111-2 reports the correlation data for comparison with the relationships hypothesized in Chapter 2. It is apparent that the find- ings are much weaker than the hypothesized relationships. In Chapter 2, 79 relationships were hypothesized to be above .4. Only 20 rho cor- relations are above that level. Another 84 are between .250 and .399, while the great bulk of the correlations are below .250. Not only are the correlations low, they are substantially different from those hy- pothesized. The proportion of correctly hypothesized relationships is abysmally low. The strengths of 278 correlations were hypothesized. 1 Only 60 (21.6%) were correctly estimated. The results conformed some- what more closely to the predicted directions of the relationships. 219 were hypothesized to be moderately or strongly related in a positive or negative direction. 131 (59.85%) are correlated in the hypothesized direction.2 In defense of these poor batting percentages, I would note ‘— II doubt that the observed proportion of correctly hypothesized vailues is different from what might be expected by chance alone. It is di fficult to support this assertion, however, because there is no way t!) determine the proportion of hypothesized values which should be as- Si'sgrmed to the categories Strong, Moderate, and Weak. For each of the ff r—sst twelve scales there are two degrees of freedom. 0n the last five, there is one degree of freedom. 2If the probability of predicting the direction of the relation- shi’ p is .5, a proportion equal to or greater than 131/219 would be pre- dic ted by chance alone in fewer than two of a thousand attempts. 108 .N.6N .m...|..61 INNMQ .6. .N. .6. .N. .6 .6 .a .N .2 :m 6.. .o. .6 .N 8:8. 3.3.8.6... .N .8 .6. .6. .N. .6. .i i .—— 60— am am am a: on aN o— fl «fl .N— a “N mmmcmmsfl 6W) LOQQJ .N .ON .6. .6. .N. .6. i .6. ... .36 .6 .m .6 .N .. 3.61.32 . .6 N 3.. Nfie. 3...... ON .6. .6. .N. .6. .6. .6. l ... .o. .6 .6 .6 .m .6 .m .. .N .N.. .m. .N .N 3 26:6; .286 ON .6. .6. .N. .6. .6. .i l .N— .—— .O— .m .w .N .m «M n— m .0 u: — .N A—IUm—OQ cm_0i_0u .N .ON .6. .6. .N. .6. .6. .—— .O— .m .m “N .m .m NJ NM . WWI—I .N— .N ANlmv Comum>L0mCOU ON .m— .m— .N— .m- om— .11 6.:— .—— 60— um am .0 am am a— “Q “N— .N a: “N ADV HCQE0>OLQE_ CQQL: .N .ON .6. .6. .N. 16H 1 .m— .+: n—— .m .0 .m 6.: .N .— .Nfl .N— .0— .w .M N ANl_v mczucmam ON .6. .6. .N. .6. .6. .i l .N. .2 .o. .6 .6 .N .m m. .N .3 .6 .6 .N E. 3.6.66: :36 00:. .....V OMN. M V I omN. H mooJ. 6.82 mmaeoz M6662 xmoz oumcoooz chLum mc_uo> __mu __o. No mucoum 0—mom 0:6 new A_Ni_v mo_nm_cm> ucuvcoaooc. ocu cuozuon oo>comno m:o_um_0ccou 0:: ”co mucoum o_mu6 mo_nm_cm> acuvcoaoa mam—am.cm> mc.uo> __mu __0. 0— can mo_nm.cm> >ucoau.umcou acoucoauuc. _N coozuon ma.cmco.um_oc wm>gmmnc: 1.N _— U _ 3.. . 109 .mu__ o_nmh mom co.ume.omc_ mum—aeoo 056 com .066; co>.m m. cm_6 mo co_umu_oc_ oc .>.0moumo xmoz ecu c. 60.nm_.m> co. oo~_6ocuoa>; No: 0.0: mcm_m omamoom .ncaom coon o>m; 6co_um_occoo u>_ummoc umcu oumo_uc_ 6.0653: o_nm_.m> oo~.o__mu. oghm .N .oN .6. .N. 6. .Nwhmnr , .6. .N. ... .6 .6 .6 .6 .N . .66.. . .6 .6 N .N-<. 666.6 e6.o.o. ON .6. .6. .N. .6. .6. 6. ii .6. .N. ... .o. .6 .6 .N .6 . .N .6 .6 .6 .N .6. no.666666 6. 6.6 ON .6. .6. .N. .6. .6. .6. ii. .6. Em........o.... N. ... .o. .6 .6 .N .6 .6 .. .N .66..6 .6 .N -moeote 6. eo.unees..o ON .6. .6. .N. .6. .6. NM .N. ... .o. .6 .6 .6 .6 .6 .. .wN_.NN_.6 .N N .6-.. 6:.6eon 66666 .N .6N .6. .6. .N. .MMN.HNN .6-.. nw— 6+: ._— em .0 6m n: .N n— .N— 6 6w 6N 6M HQOU OH CO_HMHCO_LO , . NM .6. .N. .6. ON 60— nm— nm— 6N— am a— m— «0— 6w .N. a“. 6—— 60 am a: “N ANV er—mmL Omez . ON .6. .6. .6. .6. .6. .N. .i. ... .o. .6 .6 .N .6 .6 .6 .. .N .N. .6 .N A.-6. eo.ueo.ote easemeou 00:. M V omN. H V omN. H i 600:. H 66 .6 2 M 66 Lo 2 m 6 .6 z xmo: ouncoooz mcoLum mc_uo> __mu __oc mo mocoom 0—m06 on» cam .co moLOUm o_mu6 A_Ni_. mo_nm_cw> ucovcoaovc_ usu coozuun ou>comno 6co_um_occou 03¢ mm_nm_cm> aconcoaoa to:c_ucouiim mo_nm.cm> mc_uo> __mu __0L 6. new mu_nm.cm> >ocusu.umcou ucovcuach. _N coozuun ma.cmco_um_o. wm>gmmncna. N-_: wgmgh 110 that there is little literature to guide hypothesis formulation in these areas. Social characteristics have been related to partisan voting, but much less carefully to issue orientations. The hypotheses have only served to lead into a relatively unexplored area of study. The findings below will be compared to research which was reviewed in Chapter I. It should be kept in mind that the voting scales generally reflect an orientation toward innovation, progress, change, enlarged government ' activity, and (broadly understood) liberalism. This is not to argue that each dimension of behavior lies on the same plane of a factor space with every other dimension,3 nor does it mean that the scales completely and precisely measure the extent to which the demands arising out of a dis-- tinct ideology of social liberalism have been met by the 88th Congress. However, I think most analysts would agree that several of the dimen- sions treated here are aspects of social and economic liberalism. The score values were assigned such that opposition to or non-support for the items of the dimensions would reflect low scores; the greater the support, the higher the scores. High scores on dimensions like Social welfare (J), Foreign policy (A-l), and Civil liberties (F) can certainly be regarded as central to the liberal ideology. Some of the others are more peripheral. If it is assumed for a moment that all the scales mea- sume liberalism and conservatism to some extent, it is worth looking at tive relationships of all the independent variables to these scale scores. 77113 signs of all the correlations are shown in Table III-l. In Table III I -3, the first column shows how many of the 16 correlations of each 3The relationships of the dimensions to one another is discussed br'i efly later in the dissertation. A more sophisticated analysis of this ques tion goes beyond the scope of this study. llI 6_muoh N JMN, m mcoecm... w... .HN 6697.03 .620... on; N. .ON mcoxcoz .m__ou ou.;3 N_.mf ace... c630: .m— o:_m> 0E0; cmmooz .m— p.:: mc___ozu coa meomcoa cm_oox .m— u.:: mc___uzo Log meooc cm_ooz .m— mcEEBQ __m 5.3 3.2: 6:306 N. ...: 66_c: mc___ozu no.6:ooo Loczo N..m_ ego—95c: N. .N. 0500:. cm_ouz ..— m. co_umo:oo cm_ooz .o_. N. co_umo:uo emu—_oo cu.) x 6. co..musuo .8666 66.6 66.3 N _ co_umo:oo 30. ca.) N N co_umo:oo >Lmucuso_o oum>_ca Lu.) N N can cm_oox N xoONm cm_ocom N N 0.32 N. o 6 O\O|\O —- U" [\0 LAW Ln\O\DI\\O 0% cont: Ne _ omcmzu co_um_:aom.x —wI\\O—d'm\O—d'—I\N—I\O—N—Of\ 00.... ..... v omN. H océ. H 66 66 2 M 66 66 z wee.“ 66.6.2.6.. 1603 066.060: meccum iw_occou ucoocoaooc_ o>_ummoz 0+! 2:V A II 6co_um_o.LoQ mo sumcoLum mo 2 mco.mcue_o mc_uo> .660 __o. @— ucm mo_nm_.m> acoocoaooc. _N cuozuun 6co.um_oc.oo any 60 camcocum ucm co_uoo._oii.mi___ m46o_ oocmu_m_cm_m ecu memos oo. coucoaoc oshm _m. m_. ::.m mo.w _m.m on.“ mcoecmm & oo. am.o~ mk.~ .N.“: w~.~ m_._m mtoxtoz to__oo oa_m x 00. m_.m_ m_.m mm.:: m:.w _o._: mcoxcoz cm__oo ou_;3,x oo. mm.o: _m.m_ om.m~ mm.m_ .N.:m acoc cm_co: oo. o_.w_ mm.o: om.m~_ mo.mm m~.mo_ 0:_m> 050; cm_coz .3. mm. mu. wm.~ mm. mm.~ u_c: mc___ozc con chmLoa cm_poz oo. :N.:m 0:. N_.m om. mm.: u_c: mc___ch Lou mEOOL cm_coz oo. :o.om mw.__ mo.mm m_.m_ mm.wo mc_nE:_a ..m ;u_3 mu_c: pc:0m.x oo. Nm.~o oo.m mm.mo mo.o_ ao.mm mo_ea me___ozo oo_aaooo toezo x 00. kk.m~ mm._ m~.a mm._ mo.m oo>o_aeoe= x 00. mm.~: om.mm~_ m:.__oo wo.mm~_ :m._m_m oeooc_ cm_coz oo. m~.mw No._ Jo... m~._ mm.m co_umu:co cm_c0z oo. em.Na ak.m wa.m mm.~ km.m co_oaoaoo omo__oo ;o_3 x 00. mm.mm m_.m ...ma mm.m mm.om eo_onoaoo _ootom tm_t.tb_: x 00. _m._:_ mm.~ m~.m m~.m 3N... co_umosuo 3o. ;u_3 N mo. ~m.m No.0. ...m_ mm.~_ mm.m_ co_umo:co >cmucoso_o oum>_ca ;u_3 x we. m~.m m_.m mm.m~ mm.m am.m~ omn ea_ooz ... mm.~ ma... a_.o~ m_.o_ mm.k_ xoobm ea_otod x oo. o~.mm _:.: mo.m o~.m_ o:.m_ otmoz x mm. m~._ mm._~ mo.mm oo.m~ mm.mm cant: x oo. mo.o_ mNJam mo.m~ ow.mm mm.m_ omcmzo co_um_:aom & on we o_u co_u cmoz co_u cmoz _o>04 im_umum im_>oo im_>0o mo_nm_cm> oocmo_m u atop ccmp “coccoaopc_ 1_cm_m icmum innum memo: cooZuom mcmo__n:ao¢ muncooeoo moucoLowm_o mootmcou 56mm one co muo_tun_o cmo__n:oo¢ mk— ccm muu_cum_c u_umcoosoo mmN "mo_ucma >3 mu_um_couummeu >uco:u_umcou wo cOm_cmaeouii.Mi>_ o_nmh 155 education, college education, median education, median income, owner occupancy, proportion of sound homes with all plumbing facilities, median rooms per unit, median persons per unit, median home value and median rent. Democratic district averages are higher on percent Negro, per- cent with Iow education, unemployment, and blue collar. The reported F statistic shows which mean differences are the greatest. The most substantial differences are on percent with low education and percent Negro. The standard deviations provide a basis for comparing the homo- geneity within each party on each characteristic. Looking at percent Negro, for example, not only is it true that the average proportion of Negroes in Democratic districts is higher than in Republican districts, it is also true that there is greater variation in the proportion of Negroes from district to district. Republican districts are more homo- geneous in terms of the proportions of Negroes in each. In fact, Re- publican constituencies are more homogeneous than Democratic ones on every characteristic except percent with college education and percent white collar. The data of Table lV-3 generally support Froman's (1963, pp. 90-97) contention that the parties have different kinds of con- stituencies and that these differences are associated with liberal and conservative voting. That is, chapter three showed that percentages of urban, Negro, foreign stock, low education, and unemployment were directly related to liberal voting on several dimensions. Democratic districts are significantly higher in their proportions of Negro, poorly educated, and unemployed. High school education, owner occupancy, median rooms per unit and percent farmers are negatively related to liberal voting. 156 Republican averages are significantly higher on all except percent far- mers. However, the lack of interparty constituency differences on percent urban, percent foreign stock and percent farmers makes it neces- sary to be more cautious than Froman is when he implies that the dif- ferences in the means for each party on a few factors which are ”usually associated with liberalism” explain the differences in voting by Repub- licans and Democrats. More troublesome still is the fact that the economic variables do not show strong monotonic relationships with the voting dimensions, but the parties are significantly different on each one. To look more closely at the relationships it is worth while to do a correlation analysis within each party separately. The procedure is the same as was applied using 435 districts. The 257 Democratic districts were ranked on twenty-one constituency variables and correla- ted to the roll call voting scale scores by 257 Democratic representa- tives. 178 Republican districts and representatives were analyzed the same way. If the parties respond differently to the roll call voting dimensions, then the relationships of constituency variables may be quite different within each of the two parties; that is, the proportion of high income families may be more relevant to Republican representa- tives than to Democratic ones, and the median income for Republican districts may co-vary with Republican scale scores to a greater extent than for all 435 districts and members. There are distinct differences between the correlations within each of the parties and those for all 435 districts. The complete cor- relation tables are in.Appendix F. Tables IV-4 and |V-5 summarize the 157 Table |V-4.--Strength of observed relationships between 21 independent constituency variables and 16 roll call voting variables. is correlated separatelya Each party Rho correlations observed between indepen- dent variables (1-21) and the scale scores of roll call voting--each party separately Democrats Republicans Dependent Variables Strong Moderate Strong Moderate Civil liberties (F) 2,4,5,6,1, 3,8,1_6 lO,ll,l4,l7 18.19.21 Spending (1-2) '2:“.5.11. 21 Urban improvement (D) 2,1,4,6,7, 5,8,10, 4 5,6,21 11,14,17, 12 189199.2i Conservation (B-2) 2,4,6,11, 3,5,Z,IO, 14,18,21 17,19 Foreign pOIiCY (A'l) 2911’959637: 1:8:IODJ—é ’4 2, 2991191“: 11,14,17, 17,18,19, 18.19.21 _2_1, Social welfare (J) 4,6,11 2,351, 4,6 14,17,18, 19,21 Party loyalty (K) 2,1,4,5,6, 8,10,12, 6 1,11,14,17. 1_6 18.19.21. Labor vs. business (l-l) 2,3,4,5,6, 4,5,21 .Z,ll,12,l4 17.18221 158 Table lV-4.--Strength of observed relationships between 21 independent constituency variables and 16 roll call voting variables. is correlated separatelya--Continued Each party Rho correlations observed between indepen- dent variables (1-21) and the scale scores of roll call voting--each party separately Democrats Republicans Dependent Variables Strong Moderate Moderate Agricultural policy 2: lié3§222Lt 1 _Zt__t_2 Consumer protection (B-l) 2,4,11,14, 3,5,6,Z, , l,2,4,ll, 17,18 10,19 21 14,17,18, 19,21 Negro rights (E) 2,3,4,5,6 8,16 5,6,11,15 Z,IO,11,1 I7DI8919’_ Orientation to debt (1-4) 4,6,11 2,3,5,]_, 10,14,17, 18.19.21 Space spending (1-3) 11 2,4,5,6,Z, 4,8,9,10, 14,17,18, 11,14,17, 19. 21 19 Orientation to profes- 4,6,11,14 2L1917OI9 2,516,11: 510713115111 (H) lS,2_1_ ”+217919:_2_|_ Aid to education (G) 4,5,6,1, 2,3,10,12 4,2_1 11,1 _1_6.l7.18. 19.21 aThe italicized variable numbers indicate that negative cor- relations have occurred. only weakly related to the dimension in question. Variables not reported are understood to be 159 No. m_muop m. mcoecmu x N mcoxcoz cm__oo o:_m x o mcoxcoz cm__oo mu_;3 x _ acme cm_coz _ o:_m> 08o; cm_coz m. “_c: mc___ozc con chmcoa cm_coz m “_c: mc___ozc coo meooc cm_coz _ mc_nE:_a __m ;u_3 mu_c: cc:0m x m_ mu_c: mc___ozc co_a:ooo cocZQ x o co>o_aeoc: g o osooc_ cm_cmz o co_umo:po cm_coz _ co_umo:co owe—.00 ;u_3 N o co_ecuaco _oocom cm_c cb_3 x @— co_umo:pm 3o. ;u_3 x _ N _ o o _ 4’ Ln Ch co w\ lax co_umo:co >cmucoeo_o mum>_ca ;u_3 x 0mm cm_coz xoOum cm_6cou $ _ Ocmoz & abate x _ omcmco co_um_:aom N N ONwNNMNQOCDNJOd'OLANLANONl ONMNm—QOONNOOONOOVJQLAON \ONLRNJ'NM 00:. w mmN. H. V mmN. M. WOJ. .1... M WCO_HM—OLLOU mU—flmmLm> m m0 z x w0 z m *0 z 0:1 m>_um uCOUchmUc_ A II/\ II -moz co 2 xmo) mumcocoz chLum mumcuoeoo >_oncma0m beam—Occoo m_ >ucma comm .mco_mcoe_p mc_uo> __mu __ot m. can mo_nm_cm> ucoccoaocc_ _N coozuoc mco_um_0cuoo ecu co cumc0cum ccm co_u00c_aii.mi>_ o_nmh 160 (h om~ w m. _— m— _— m_ m— _— o— m— o— :— m— :_ m— m N. w m— _— :— m_mu0h mcoecmu N mcoxcoz um__oo o:_m x mcoxcoz cm__oo 0u_£3 & acme cm_poz o:_m> meo; cm_coz u_c: mc___ozc Lou mcomcoa cm_poz u_c: mc___ozc Lou mEOOc cm_coz mc_nE:_a __m Lu_3 mu_c: pc:Om x mu_c: mc___03c co_a:uuo cocZO x co>o_aeoc: N meooc_ cm_poz co_umo:co cm_cmz co_umoauo moo—_oo ;u_3 x cotocoabu _oocum ;m_; cd_3 x co_umu:co 30. :u_3 x co_umo:vo >cmucoso_o mum>_ca cu_3 N can cm_coz xoopm cm_0com N ocmoz x :85 .N. omcmco co_um_:aom N — Nmomqmommmmooq—ommmowl?‘ omomoooooo—oo—oooooomlm G\W\¢>OJOIN\N\OIOJOJOJF~C>Od4r4rCVAT¢Vr\rn 00:. H mmN. H V mmm. H mco_um_0ccoQ a co 2 a co 2 a to z oce u>_bb /\ 11A 11 mo 0 41- + i /\ II mo_nm_cm> ucoccoaonc_ imo xmo: mumcocoz chLum z wo z mcmo__naaom po:c_ucou >_0umcma0m beam—mucoo m_ >ucma comm .mco_mcoe_p mc_uo> __mo __oc m. can mo_nm_cm> ucoccoaocc_ _N coozuon mco_um_0ccou ocu mo sumCOLum pcm co_u00c_aii.mi>_ o.nm# l6l findings and can be compared to Tables 111-2 and 111-3 for all districts. One difference involves the signs of the correlations between the educa- tion variables and the roll call voting dimensions. In the hypotheses of chapter two I expected in general that the higher the education variables, the more liberal congressional voting would be. The rela- tionships reported in chapter three are generally moderate to weak, but the direction of relationships were opposite to that hypothesized. (Examined within parties, however, it is generally true that the higher the education variables for the districts, the more liberal the voting of the representatives from those districts. The correlations remain low, however, and this statement is more relevant to Democrats and Demo- cratically controlled districts than it is for Republican ones. Another striking sign change occurs with the correlations between the proportion of Negroes and the policy dimensions. Unseparated by party the correla- tions with the proportion of Negroes were modest but mostly positive. Among Democratic districts and Democratic representatives all the corre- lations are negative. Among Republicans there are a few more negative relationships but the correlation coefficients are all below t .250. The explanation here is simple. Southern Democrats, with the highest propor- tions of Negroes, vote more conservatively than their northern and wes- tern colleagues--a finding consistent with those reported by journalists and academics alike. However, the contrast in findings between all dis- tricts and Democratic districts hints that Republicans vote more conser- vatively than southern Democrats. This suggestion will be explored further below. There are striking differences between the two parties which are evident in Tables IV-4 and |V-5. Weak correlations remain most numerous, 162 but the moderate and strong relationships together outnumber the weak ones among Democrats. In fact, the findings among Democrats compare more favorably to the values hypothesized in chapter two than the find- ings of chapter three for all members. For Republicans, however, con- stituency variables are only mildly related to roll call voting. For Democrats economic variables are among the strongest correlates on the fourteen scales where relationships are 2 f .250. In every case the associations are positive; the higher the economic well-being in Demo- cratic districts, the more liberal the roll call voting by the repre- sentatives tends to be. Among Republicans the pattern is much less distinct. There is more support among Republican Congressmen from economically higher ranking districts on Foreign policy (A-l), Consumer protection (B-l), and Space spending (1-3). The tendency is weak, but similar on Negro rights (E). The pattern among Democrats is much clearer on social policy dimensions than it is on economic policy dimen- sions. Given the correlations among Republicans and Republican districts and all representatives with all districts, the correlation coefficients found among Democrats and Democratic district characteristics are very large. Median income relates to Civil liberties, rS = .671; to Urban improvement, rS - .626; to Social welfare, R5 = .404; to Negro rights, r5 - .669; to Consumer protection, rs = .455; and to Aid to education, r5 - .417. The education variables are not much more strongly related to the voting dimensions among Democrats than they are for all districts and members. The associations among Republican and Republican district characteristics are weaker still with one exception. I hypothesized strong relationships between education variables and voting and I was 163 particularly interested in the proportion of college educated and roll call voting. The relationships with this variable have been minute to the point of insignificance. It does appear above the .250 level three times when Republicans are taken separately; it is positively related with Space spending and Foreign policy, and negatively with Agricultural policy. Percent college does not discriminate among Democrats. Low education works the other way around. It is moderately to strongly negatively related to all the social policy dimensions, as well as Conservation, Foreign policy, Party loyalty, Labor vs. business, Orien- tation to debt, Space spending, and Orientation to professionalism. This variable does not discriminate among Republicans in roll call voting. The variable most consistently and strongly related to liberal voting among Democrats is the proportion of Foreign stock in the dis- tricts. The highest correlation of constituency characteristics and voting reported in chapter three was between Foreign stock and Negro rights (r5 = .643). Among Democrats alone that correlation is rs = .767, again the highest of all the correlations between characteristics of Democratic districts and the dimensions of voting. 0f the fourteen dimen- sions on which there are correlates higher than i .250, the highest cor- relation is with the percent Foreign stock for nine scales. Four other scales relate more strongly to the percent with private elementary educa- tion which, as was previously noted, is the independent variable most similar to Foreign stock. This variable also relates more strongly to the dimensions of social policy than it does to economic policy dimen- sions. Percent Foreign stock is not as consistently the high correlate of voting dimensions for Republican districts and representatives. It 164 is the most highly correlated variable on six of the eleven dimensions which have relationships higher than t .250. The highest correlation in the matrix for Republicans only is between Foreign Stock and Orien- tation to professionalism (r5 = .521). The tendency to relate more strongly to social than economic variables is not statistically signifi- cant among Republicans.“ Percent urban and percent farmers continue to covary with policy dimensions but not as closely within the parties as over the 435 districts and representatives. The divergence is greater among Republicans than Democrats. For Democrats the relationships between these variables and the scale scores for each dimension are generally stronger than when 435 districts and representatives are treated. Percent urban correlates with liberal voting especially on social policy dimensions, but also strongly on Conservation, Foreign policy and Party loyalty. Percent farmers includes a strong negative relationship with Orientation to professionalism. Among Republicans, however, percent urban correlates less strongly than it does using all districts. This tends to be the case with percent farmers, except in a couple of spots. There is a strong negative relationship between farmers and Space spend- ing. This probably identifies the opposition of representatives from rural areas who prefer space expenditures for defense rather than lunar research. Among Republican districts the proportion of farmers is strongly positively correlated with liberal voting on Agricultural policy (C). At first glance it seems incongruous that the correlations “The statistical test is the Mann-Whitney_U Test used as it was in chapter three and presented in Appendix E. For Democrats the dif- ference is significant at the .01 level of significance. 165 overall and among Democrats should be negative for this relationship while the coefficient for Republicans is strongly positive, while at the same time the parties are strongly different in that Democrats support Agricultural policy to a greater extent than Republicans. The explanation is in the distribution of scale scores. Democratic sup- port is concentrated at the extreme liberal end of the Agricultural policy dimension. This support comes not only from districts with sizeable farm populations, but from highly urban districts as well. Republicans are more evenly distributed over the dimension, but are concentrated on the right. The correlation reflects the slightly more liberal voting by representatives of the midwestern farm districts, which nevertheless remains considerably to the right of the votes by the Democrats. Median age within a district is not a very powerful correlate of roll call voting, but it relates more strongly among Democratic dis- tricts and Democratic voting than it does among Republican or overall. I think the explanation here is more in the nature of this constituency characteristic than the voting behavior. Among Democrats median age covaries with urbanism, and thus with liberal voting. Among Republicans this covariance is much lower, probably because in Republican districts the small towns are havens for the elderly. Nevertheless, contrary to the original hypotheses, there is not a direct relationship between age and conservative voting, not even when considered among Republican dis- tricts only. Home ownership has long been thought to be an important political variable. As mentioned previously MacRae (1952), Froman (1964), and Wilson and Banfield (1964) found it to be related to policy decisions 166 both in representative voting and referenda. It is not highly correlated with other independent variables such as median family income, percent with college education or percent white collar, but chapter three showed that it was consistently negative as it is correlated with voting on the policy dimensions and most of these relationships were of moder- ate strength. Separation by party makes a notable change. Within the parties this variable only weakly discriminates policy voting. No such transformation occurs with any other variable. The meaning is simply that percent owner occupancy is related to which party controls the dis- tricts; the higher the home ownership, the more likely the district is to be Republican.5 Looking back to the all districts correlations of chapter three (Table III-l), it can be seen that higher negative correla- tions are with the policy dimensions which evidence the strongest party cleavage (Table IV-l). This is confirmed by a rho correlation of the two sets of ranks: r5 = .839. The finding here is different than MacRae's (1952). Roll call voting within the parties does not vary with the extent of home ownership. In neither party is there a relation- ship between owner occupancy and any of the sixteen dimensions of voting which is significantly different from zero (P = .001). Percent white collar is generally more strongly related to the policy dimensions within the two parties separately than when the over- all relationships are taken. Overall white collar correlated weakly with fifteen of sixteen dimensions. The correlations for each party separately are moderately or strongly positive on four dimensions: 5The mean home ownership for 178 Republican districts is 58.04 percent (standard deviation = 9.00); for 257 Democratic districts it is 68.59 (standard deviation = 16.08). For all 435 districts the mean is 62.36 percent (standard deviation = 14.58). 167 Consumer protection, Orientation to professionalism, Foreign policy, and Space spending. Why the correlations go up in both parties on these dimensions, 1 cannot explain. Among Republicans there is co- variance between the correlations of percent college and percent white collar, but not among Democrats. However, among Democrats percent white collar also correlates more strongly with Civil liberties, Urban improve- ment, Social welfare, Party loyalty, Negro rights, Orientation to debt, and Aid to education. Among Democrats white collar generally correlates with more liberal voting. The only dimension on which the Republican correlation is substantially stronger is on Agricultural policy. Per- cent white collar is negatively related to Agricultural policy for Republicans: r5 = -.393. For Democrats the correlation with Agricul- tural policy is not significantly different from zero. The median number of persons per occupied unit, has not re- ceived much attention. I noted in chapter two that because it is nega- tively associated with median age, it probably gives evidence of young families. It is somewhat related to home ownership (see Appendix C). Overall this variable is only weakly related to policy dimensions. Among Republicans it does not discriminate according to voting, but it does pattern with Democratic voting. There are modest negative rela- tionships with Civil liberties, Foreign policy, Party loyalty, Negro rights, and Aid to education. 1 can offer only a very tentative ex- planation for this pattern. I do not think these relationships are simply covariates of the associations of these dimensions with other variables. Five other variables in the matrix of Democratic correla- tions are mostly negative. None regularly covaries with median persons 168 per unit. I think this variable detects Democratic districts where there is a prominance of young families, probably of modest income, who are or who look forward to becoming home owners. They probably fit the ”marginal homeowner'I description. If such a pattern is per- ceived in the representational process, it could support a sense of moderation for liberal voting. Several variables simply do not correlate with the voting patterns in any substantial way. Population change, median rooms per dwelling unit, and percent blue collar do not correlate with voting overall or within the parties separately. Percent unemployed does somewhat within the Democratic districts, but this variable is more strongly related to districts and roll call voting overall than among Democrats only. I have left percent Negro as the final variable for considera- tion. Among Republican districts and representatives, none of the correlations are significantly different from zero (P I .001), but for Democrats all but two (Agricultural policy and Foreign trade) are. All these significant relationships are negative. The reason is obvious, of course. Percent Negro generally shows the tendency of southern Democratic representatives to vote in a less liberal fashion than their northern counterparts. This is the evidence of the correlations, and these merely give support to conventional wisdom. What is more interest- ing is that these correlations suggest an order of deviation by the southern representatives from the northern majority of the party. The greatest deviation is on the Negro rights dimension (E) and the least is on Agricultural policy (C). I will examine the extent of cleavage 169 between the southern and nonsouthern representatives in chapter five. Looking once more at Tables IV-4 and IV-5, it is apparent that the roll call voting behavior of Democratic representatives is more Strongly associated with constituency differences within the party than is the case for the Republicans. This is not explained away simply by the fact that there is more homogeneity among Republican districts than Democratic ones. This homogeneity is of some effect, of course. The low mean and standard deviation on percent Negro (see Table IV-3) is probably the major reason why this variable is not monotonically related to voting on any of the policy dimensions. Similarly small variation may account for the modest correlations which occur with median age, percent with poor education, percent with college education, median education, unemployment, median rooms per dwelling unit, and median persons per dwelling unit. There are a few variables which, in spite of considerable variability, do not relate to scale scores of voting in either party, such as percent population change, owner occu- pancy, and blue collar workers. Even making these exclusions from a comparison of the two parties, ten variables remain which invite com- parison. There are 160 correlations (10 X 16). In the Republican matrix 106 are weak, 45 moderate and only 9 which are strong. For Demo- crats 33 are weak, 50 moderate, and 77 are strong. Examined variable by variable, the Republican matrix has more weak correlates and fewer strong correlates in every case. If the constituency influence theory does causally relate to roll call voting, then the data of this chapter indicate that the impact is made more strongly upon Democratic representatives than upon 170 Republicans. This tentative conclusion should be accompanied by some hedging, of course. fIt only holds given the data of the analysis. Other conclusions might emerge with a different selection of constit- uency variables. The patterns are by no means deep and irreversible. Nevertheless, they are consistent. 0n almost every dimension of be- havior the pattern of relationships is stronger between Democratic voting and the characteristics of Democratic districts than Republican voting is with the characteristics of those districts. Narrowing attention to a few variables, as suggested in the last paragraph, the variables most related to voting become apparent. Percent foreign stock dominates the covarying pair, foreign stock and percent with pri- vate elementary education. Median family income is representative of the economic variables. Percent urban and percent farmers are very similar for Democrats, but measure somewhat different things for Repub- licans. Percent white collar and percent with high school education remain, the latter has the most comparable measurement properties when looking at the two parties. Looking at the strengths of the correla- tions, Foreign stock ranks highest within each party. Ranking the correlations Of these six variables has been.done in Table.IV-6. Rank- ings are not made on dimensions which have only weak correlations with all six of these variables. (The correlations themselves are reported in Appendix F.) Table IV-6 suggests a couple of things. One is the stability of the relationships among the Democrats. The order of strength for the relationships of the independent variables with the dimensions is very consistent where there are correlations of at least moderate size. Foreign stock is most related to the scales of voting. 171 Table IV-6.--Ranks of the Rho correlations for six independent vari- ables over the 16 dimensions of roll call voting Democrats C o 73 m U 8 2 ° :2 ...- L 8 2 Dependent '3 '5 L Variables 8 m g .9 in ..c O — c .9 8 8 v. 0'1 .1: -— L c .... 0 0 m 0 .c c u E .0 L H in .. 1.. s .2 'E g s :2 3‘2 3‘2 3‘9 Z N N Civil liberties (F)a 3 1 6 2 5 4 Spending (1-2)a 3 1 6 4 5 2 Urban improvement (0) 4 1 6 2 5 3 Conservation (B-2)a 4 1 6 2 5 3 Foreign policy (A-l) 4 1 6 2 5 3 Social welfare (J) 4 1 6 2 5 3 Party loyalty (K)a 3 1 6 2 5 4 Labor vs. business (1-1) 3 I 6 2 5 4 Agricultural policy (0) O O O o 0 0 Consumer protection (B-1) 3 1 4 2 6 5 Negro rights (E) 4 1 6 2 5 3 Orientation to debt (1-4)a 4 l 6 2 5 3 Space spending (1-3) 3 2 6 l 5 4 Orientation to rof— sionalism (H) 4 2 6 1 5 3 Aid to education (6) 3 l 6 2 5 4 Foreign trade (A-2)a O O 0 O O O 2 ranks 49 16 82 28 71 41 Median ranks 3.5 1 6 i 2 5 3 172 Table IV-6.--Ranks of the Rho correlations for six independent vari- ables over the 16 dimensions of roll call voting--Continued Republicans V! L 0 x — L. 8 3 t. s .. Dependent ‘3 m- g _9 Variables m f; 8 '5 c ...- C C U V1 01 .C O -— 1. c .— o— O) 0 1‘0 0 .l: u c: u E '° L :3.“ .3 IE 3 5' t9 3 1:‘3’ '3 3 u. 3‘2 3‘2 82") Z 3‘9 SQ Civil liberties (F)a o O O o O 0 Spending (1-2)8 0 o o o o 0 Urban improvement (D) 3 1 6 4 5 2 Conservation (B-2)a 0 0 0 O 0 0 Foreign policy (A-l) 5 1 6 2 3 4 Social welfare (J) 4 l 6 5 3 2 Party loyalty (K)a O O O O O 0 Labor vs. business (1-1) 4 l 6 5 3 2 Agricultural policy (C) 3 6 5 2 4 1 Consumer protection (B-1) 4 l 6 3 5 2 Negro rights (E) 5 3 6 1 4 2 Orientation to debt (1-4)a 0 o O 0 O 0 Space spending (1-3) 2 4 6 5 3 1 Orientation to profes- sionalism (H) 5 l 6 3 4 2 Aid to education (G) 5 l 6 4 3 2 Foreign trade (A-z)a O o o o o O 2 ranks 4O 20 59 34 37 20 Median ranks 4 l 6 3.5 3.4 2 8If in either party none of the six independent variables cor- related with a dimension rS 2 t .250, ranks were not calculated. 173 The higher the proportion of foreign stock, the more liberal the voting. Income is next. Percent urban and percent farmers covary closely any- way so their similar ranks are not surprising; of course, percent urban is directly related to liberal voting and farmers is negatively related to liberal voting. White collar and high-school education trail behind. This similarity of ranks says something about the scales dimen- sions too. The correlations between the scale scores is quite high. Perhaps there is an underlying dimension for which these are simply variations. The criteria of scale analysis may be too demanding a set of criteria to establish the content of this dimension, but much of the voting, particularly among Democrats, may fit a simpler left-right con- ception.6 The pattern is much less clear for Republicans. For one thing, of course, the correlations are smaller. Percent farmers is noticeably more related to the dimensions of behavior among Republicans than among Democrats. Surprisingly, percent foreign stock, though the relationships are less strong than for Democrats, is the variable which ranks highest) on more dimensions than any other. It is directly related with scale scores, indicating that there is a tendency for more liberal voting by 6During the scaling process I attempted to scale 18 roll call votes which were used in the Congressional uarterl as "Larger Federal Role“ items. They did not produce an acceptable scale. CR = .882 and CS - .706. According to the item criteria I used, several items would have had to be excluded if one wanted to develop a scale with acceptable reproducibility. I do not doubt that several of the items in the scale would go together in such a fashion as to produce acceptable coefficients of reproducibility and scalability. However, if ngs item selection is to be used to test whether “Larger Federal Role” is a dimension by the criteria of Guttman scaling, those standards were not met in the 88th Congress. Elementary linkage analysis (McQuitty, 1957) is a technique for isolating types ”in which every member of a type is more like some other member of that type (with respect to the data analyzed) than he is like 174 Republican representatives from districts with larger proportions of first and second generation foreign stock. It is interesting that percent foreign Stock and percent farmers should be the clearest cor- relates of voting. This seems to confirm the conventional view that native, small town folk are basis for social and economic conservatism. It is unfortunate that percent urban is such a blunt measure, not dis- tinguishing small town urbanism from moderate and large city urbanism. any member of any other type.” I applied this technique to a matrix of correlations of the scale scores. The intercorrelations are generally higher among Democrats than among Republicans or among all representa- tives. However, using 257 Democrats 3 types emerged. CSMR PRO_ S6\\‘M 82 NEGRO RTS : URB IMP ../ CIV LIBS 61/71 ‘N\74 CONSERV FRGN POL. Type I /; PRO-ISM SPACE $ \68 52/ 73 55 69 PRTY Lov4—.___—_/.—AIO souc $$$$$$ éDEBTa—T—‘SOC I«Nil-F 2843 66)\\ AGRIC LABOR 29 FROM TRD Type II Type III 175 The variation in the rankings gives more evidence that voting patterns are more complex among Republicans than Democrats. Median family income relates moderately to liberal voting on Negro rights, but its relationships with other social and economic dimensions are much more modest. All the relationships among these six characteris- tics of districts and roll call voting are weak on six dimensions. This is 662 because of invariant voting and high agreement among Repub- licans. To the contrary, as Table lV-2 has previously demonstrated, the lower standard deviation coefficients on the Republican side indicate that Republican scale scores are spread more evenly along the dimensions than is the case for Democrats. The mean for this standard deviation is the mean frequency for each scale type on a given scale. Republicans more closely approximate these means than do Democrats. Democrats tend to be unevenly distributed, often bunched together on the extreme left end of the scale dimension. This bunching is evidenced by the higher standard deviation coefficients. Despite this bunching effect among Democrats compared to the more even distributions for Republicans, the voting variation is regularly correlated with constituency characteris- tics. Among Republicans this is not true even where, in terms of ---4=*- A reciprocal pair of variables; each is the other's highest 7 O O correlate in the matrix. -—-—-——€> The variable at the tail of the arrow has as its highest correlate the one at the head, but the one at the head is higher with some other. In the original matrix all correlations were to 4 places; here they are rounded to two. The decimal point has been dropped. This technique demonstrates the existance of at least three types among sixteen dimensions. 176 measurement criteria, the scale discrimination has been very good--on dimensions including Civil liberties, Spending, and Conservation. Summary The studies cited in the beginning of this chapter which Show that system outputs are more strongly related to social and economic traits of the systems than to the levels of partisan conflict within them ought not be interpreted to mean that intra-system party conflict is irrelevant to the well-being of system members. The important re- sults of party conflict is ”who gets what”--who is to receive benefits of public policies and who will pay for these benefits. In chapter one I hypothesized that ”who gets what” could be explained in large part without reference to parties and that party identifications simply over- lay and conceal constituency differences which account for differences in the policy voting of representatives. Even the modest levels of correlation which I hypothesized would occur in the findings were too demanding. The expected relationships simply did not materialize. This chapter has centered on the partisan conflict of roll call voting in the House of Representatives. It shows that cleavage is generally substantial, but that there is a good deal of variation in the extent of party cleavage among the sixteen dimensions under examina- tion. The sharpest differences between the parties are in the way they vote on agricultural and economic policy dimensions (ignoring Party loyalty, which includes several procedural items). Conflict is moder- ate to slight on social policy dimensions. Foreign policy conflict is moderate; broad bipartisan unity cannot be taken for granted. 0n the dimension most associated with drama and public concern, Negro rights, 177 the parties as such were not in conflict. The Republicans were generally less cohesive in their voting than Democrats. The major exception to this generalization being that Democrats show greater variation on Civil liberties voting. This find- ing should be taken with some caution, however. It will be recalled from chapter two that the scale scores tend to be skewed to the left on most of the scales. Part of the unity of the liberal voting is, I think, and artifact of the function of the committee system. Very often the difficulty for social and econOmic liberals is getting a pro- posal to the floor for a vote (see, for example, Fenno, 1962; Robinson, 1964). What Fenno refers to as the negative verb actions (”cut,” ll ”wring,” ”trim,” ”carve,” ”slice,” ”prune,” “whittle,” ”squeeze, ”lop off,” ”chop,“ ”slash,” ”pare,” ”shave,” and “whack“) have already taken place by the time a bill comes to a record vote. The liberals then, generally unable to get the whole loaf, vote in a unified way, with their fellow partisans to accept the half loaf. Discriminating the degrees of liberal voting with scale analysis requires more liberal proposals getting to the floor to be voted up or down. Testing this notion goes beyond the scope of this inquiry, but I think it can be done by looking at voting on bills reported out by committees containing disproportionate numbers of liberal members. It is likely for example that unity among U. S. Senate liberals of the 88th Congress is lower on foreign policy votes than on finance bills because of the differences in the character of the proposals which emerged from Fulbright's com- mittee compared to those from Senator Byrd's. It may be that the vari- ation on Civil liberties for the House in the 88th Congress roll call votes reflects upon the character of the membership of Representative 178 Celler's Judiciary Committee and the relatively more liberal proposals which were reported to the floor (in spite of delays in Judge Smith's Rules Committee). The constituencies of Republican controlled districts are sub- stantially different from those of Democrats. It is difficult to directly attach these differences to differences in voting by the two parties. Variations in voting within the parties can be associated with variations in constituency characteristics. There is generally higher covariation in constituency characteristics.and voting among Democrats than among Republicans. For Democrats several variables relate directly with liberal voting on most of the dimensions of voting behavior: percent urban, percent foreign stock, percent private elemen- tary education, percent with less than five years of school, median income, percent sound homes with all plumbing, median home value, and median rent. Percent farmers is negatively related. Two variables consistently rank highest on these dimensions. They are median income and foreign stock, and not surprisingly, these two independent variables are highly related to one another (rs = .775). The variation in Republican voting is generally much less cor- related to the constituency variables used in this study. Not only are the correlations lower, the ranks of the correlations of the inde- pendent variables with the voting dimensions vary more than among Democrats. Percent foreign stock is the variable usually most related to liberal voting by Republicans, but the relationships are much weaker than among Democrats. Median family income and the other economic indicators are substantially less related to variation in Republican 179 voting than they are to variations in Democratic voting. The other variable which has high ranking correlations with Republican voting is percent farmers. The signs of these correlations are nearly all nega- tive £§££E£ on the Agricultural policy dimension. This correlation, the second highest in the whole Republican matrix, is a positive .479. These findings coincide closely with Mayhew's (1966, p. 40 and m), who says, ”Over the sixteen-year period (1947-1962), the more farmers a Republican Congressman represented the more disposed he was to desert his party on farm votes. Heresy flourished in the heavily agricultural districts of the Corn Belt as well as in the poorer dairy districts and in the wheat areas of the plains.” The surprising finding to me is the fact that the variation in Republican voting on several dimensions is associated with none of the variables under examination. The data simply do not reveal regular monotonic relationships on what are very substantial dimensions of cleavage between the parties--especia11y Party loyalty, Orientation to debt, Spending, and Civil liberties. What remains for this study is the examination of a number of political variables. If party is very relevant to roll call voting other aspects of the partisan situation, region, competitiveness, ambi- tion for higher office, and reelection may be related to roll call voting. These variables are examined in chapter five. CHAPTER V INTRA-PARTY PATTERNS IN ROLL CALL VOTING Some aspects of the constituency are difficult to render into hard and intersubjectively meaningful indices. In chapter one I noted that many citizens hold the view that congressmen ought to vote ”the way people feel.” Certainly many congressmen take pains to act in what they feel are the best interests of those in their home districts (see Dexter, I957; Wahlke _1 _1. 1962). Wahlke _1 _1. treated role orien- tations of legislators as an important variable in legislative behavior. Similarly the orientation of constituents toward representation are probably different in different constituencies. Certainly their policy preferences are different. Doubtless some have positive preferences-- new programs of government ought to be enacted; others seek to prevent initiation and change. There may be variables more basic than character and extent of policy demands. Constituency attentiveness may be an im- portant variable. How interested, concerned, and informed are the con- stituents? It is known, of course, that generally attention and infor- mation are at very low levels (Stokes and Miller, 1962), but how important is this as a variable, both over time, and across the constituencies of the Congress? Citizen sense of efficacy is another variable which may affect the representational process. It affects electoral turnout, but does it structure the expectations that constituents have of their 180 181 representatives? What are the expectations of constituents? Do they want to be consulted by their representatives? Do they expect their representatives to estimate and act in their interests in a prescribed manner, or do they prefer representation of a ”public-regardingness” view? These and many other questions need to be asked of constituents and asked of their representatives. Representatives' perceptions of constituent opinions have proved to be significant variables in the pro- cess of representation (see Miller and Stokes, 1963). The point of these questions is to emphasize that it is doubtless important that more be known about the nature and effect of political culture upon politics generally and representation in particular. Region There is a good deal of intuitive analysis based upon cultural interpretation. David Potter (1954, p. 67) argues that the ”. . . state of relative abundance, of material plenty, has been a basic condition of American life and that it has had a pervasive, if undefined, influence upon the American people.” Yet not all parts of the system share identically in this abundance. Particular subsets of the society are relatively deprived. The great urban ghettoes of this country have always contained the potential for politician initiative. Different ethnic groups have ”seen the opportunities, and took 'em,” to paraphrase Boss Plunkitt's words, or failed to take them, as the case may be (see, for example, Cornwell, I964). Charges of “bossism” have more conse- quence in some communities than in others (Riedel, I964), The cultural and political implications of race are different in different parts of this country. Indeed, the values and expectations upon which the 182 political processes of the south have been built cannot be understood without an understanding of the cultural and political meaning of race (Key. 1949). Testing the implications of political cultural variables upon patterns of representation in the 88th Congress goes far beyond the 'scope of this study. Nevertheless, the study provides some potentially valuable data for such more complex inquiries. I have produced measures of individual performance by representatives on a number of dimensions of roll call voting. Several, if not all, of these dimensions can be expected to have continued existance and importance in future congresses. It is well, therefore, to provide some elementary description of roll call voting predilections. Because such voting is so strongly grounded in the partisan context, the distinctions to be made are within the re- spective parties. Instead of precise analytical distinctions in political culture, I will simply make regional comparisons. The most apparent and broadly advertised intra-party cleavage in Congress is that between southern and northern Democrats. While a regional definition based upon the boundaries of a cluster of states seems to be a crude way of marking off a deviating political subculture, there are both justifications and pre- cedents which rationalize what happens to be a convenient delineation. There is no need to thoroughly recount the events which established and perpetuate the unique political orientations and partisan attachments in the South (for a concise summary, see Goldman, 1966, pp. 133-136). Suf- fice it to say that economic rivalry, war, reconstruction, and racism resulted in quite a different ”natural order" of things among southern 183 Democrats than exists among their nonsouthern fellow partisans. It is true that the solidarity of the southern states has changed in the recent period, particularly in presidential elections, Probably, however, the best evidence of the remaining southern solidarity appears in the pat- terns of congressional representation. Different scholars define the South in different ways. Key (1949) studied the eleven states which formed the Confederacy. Conqressional Quarterlx adds the states of Kentucky and Tennessee to this group. My own usage is based upon the definition of the Survey Research Center. Ten states are named as Southern: Alabama, Arkansas, Florida, Georgia, Lousiana, Mississippi, North Carolina, South Carolina, Texas, and Virginia. These ten States contained 97 congressional districts. For the 88th Congress only eight had Republican representatives (Florida 11 and 12; North Carolina 8 and 9; Texas 5 and 16; and Virginia 6 and 10). Thus, for purposes of this analysis there were 89 southern Democratic districts. The remaining 168 nonsouthern Democratic districts are used for comparison. The intra-party differences in roll call voting are manifest in Table V-l. The method of analysis is the same as that used previously to Show the differences between the parties in roll call voting and can be compared to Table IV-l of the preceding chapter. H coefficients greater than 10.837 indicate statistically significant differences (P . .001) between the two groups. The dimensions are presented in the order of increasing differences between these two regional wings of the Democratic party. The data confirm the expectation that there are dif- ferences between the southern and nonsouthern groups of democratic repre- sentatives. 184 Table V-1.--The degree of differentiation between southern and nonsouthern Democrats on 16 dimensions of roll call vot- ing (Kruskal-Wallis H Coefficients) Foreign trade . . ... . . . . . . . . . . . . . . . .50 Agricultural policy . . . . . . . . . . . . . . . . 13.05 Consumer protection . . . . . . . . . . . . . . . . 35.10 Space spending . . . . . . . . . . . . . . . . . . 35.31 Spending . . . . . . . . . . . . . . . . . . . . . 40.67 Orientation to professionalism . . . . . . . ... . 55.22 Labor vs business . . . . . . . . . . . . . . . . . 63.10 Orientation to debt . . . . . . . . . . . . . . . . 63.80 Conservation . . . . . . . . . . . . . . . . . . . 70.68 Aid to education . . . . . . . . . . . . . . . . . 86.71 Social welfare . . . . . . . . . . . . . . . . . . 87.04 Foreign policy . . . . . . . . . . . . . . . . . . 100.67 Party loyalty . . . . . . . . . . . . . . . . . . 108.56 Civil liberties . . . . . . . . . . . . . . . . . . 125.11 Urban improvement . . . . . . . . . . . . . . . . . 125.60 Negro rights . . . . . . . . . . . . . . . . . . . 168.65 As one would hypothesize, the sharpest internal cleavage between South and nonsouth is on the Negro rights dimension. Support for Negro rights is very much higher among nonsouthern representatives than among southern ones. No southern representative received the highest possible score on this scale and only one nonsouthern Democrat (Everett, Tennessee 8) received the lowest possible score, opposing the Negro rights proposal which received the highest proportion of support (84.1% of those who participated; 1963:33. See chapter two and Appendix B) of all Negro rights proposals which came to a roll call vote in the House. Meanwhile 104 nonsouthern Democrats (over 41% of those scaled) received the highest possible scale score, and 44 southern Democrats (almost 18% of all the Democrats scaled) received the lowest scale score. Table V-l also shows that the intra-party conflict among Demo- crats is not limited to Negro rights alone. There are statistically significant differences between the southern and nonsouthern wings of the party on every scale dimension except foreign trade. On every dimension 185 it can be Shown that the rankings of support by nonsouthern Democrats I Interestingly, the evi- are higher than those of southern Democrats. dence of the rankings is that there is much more intra-party conflict on the dimensions of social policy than there is on the economic policy dimensions. Although the cleavage is not inconsiderable, voting on the economic policy dimensions does not arouse the southern opposition that Civil liberties, Urban improvement and even Conservation roll calls do. In spite of the fact that organized labor has suffered many setbacks in the South (see, for example, Barbash, 1958, pp. 42-44), the southern representatives do not oppose the bread and butter issues of the Labor versus business dimension with the intensity that they oppose social policies often endorsed by organized labor unions. Similarly, conflict on spending items is mild compared to that on social policies, Party loyalty and Foreign policy. The latter two dimensions receive consider- able southern opposition. -Agricu1tura1 policy, on the other hand, gets almost as much liberal support from the southern delegations as from any other region. However, among the few Democratic opponents of liberal agricultural policies are several southerners, particularly from Florida, who opposed all the items of the Agricultural policy dimension. AThe constituency data of Table V-2 are very interesting. The correlations between the constituency characteristics of Democratic dis- tricts and the voting by Democratic representatives are largely lTo produce the KruskaI-Wallis H coefficients the scale scores are ranked from one to N, with N equal to the number of representatives scored on the scale. The sums of the ranks for each group is produced. The computer program used (Morris, 1966) reports the rank sums and the number of observations for each group. The average rank for the group can easily be determined indicating which group, as such, had the higher average ranking and scale scores. 186 .moo. cmnu mmo. m. .o>o. oocmu.w.cm.m on» umzu mOumu.oc. 00. oeucoaoL ochm oo. m~.N: m~.0 Nm.~. mm.m _m.m ntoetme x mm. _o._ as.“ N:.om mm.“ om._m atoxtos Ln:oo us.m x oo. mm.cm ...“ mm.cm _m.m m_.m: ntoxtoz tn__oo ou_c2 x oo. om..__ _c.o_ ~m.~m ~_.:_ mm.o~ acct cn_ooz oo. mo.~k N:._~ mm.mm ~a.~m mm._~_ ua_n> oeoe cn_coz oo. ec.m~ mu. mo.m em. kw.~ o_ca mc__.ozo cue neontoa cn_ooz mu. mm._ :N. ac.: mm. .N.: o_ca mc___oso toe neoot cn_euz oo. ...mw o~.N_ m_.wm mm.m_ m~.:~ mc_bEa_o __b to_3 nu_ca ecaom x No. me.m om.o _m._c ~_.m_ _m.cm no_ca m5:26 co_eauuo tocso x 00. mm.m_ oo._ mm.: wo.~ No.6 oo>o_eeoc: x oo. o~.ma_ oo.mcm m_.aeoa mm.:___ a~.-~m oeouc_ cn_ooz oo. ow.~a o... _m.m N... mm.o. co_onuaeo cn_ooz mm. am. 00.. mm.c km.~ No.6 co_otuaco omu__ou tb_: x 00. oo.m~ om.c __.mm -.m Mk.mm co_unuaao _oocum cm_e to.) x 00. N_._m_ -.c ea.c_ c_.a ca.m co_unuaco so. to.) x 00. m~.ao_ mm.m mc.a om.~_ mm.w_ eo_onoaeo stepco5u_u oub>_te ep_: x oo. em.me m~.m Ne.c~ mm.m -.om omn cn_eoz oo. ~_.om_ mm.c mm.: om.m_ -.a~ xuoun cm_otoe x 00. ~:.mm :m.~. 0m.:~ 03.3. mm.0. oLmoz N 00. mm.um om.o~ 0_.mm oo.m~ No.~m cont: x mm. 00. ::.0~ m:.m. m...: mm.m. omcmzo co.um.:aom & mm o.u co.u coo: co.u com: 00 .o>04 im.umum im.>00 im.>00 mo.nm.cm> ucoocoaooc. oocmo m ten new -_c_cm_m -ccnem -ocnum memo: coozuom mumcooeoo mumLUOEoo oucoLomm.0 cLOLHJOm ccocu30mcoz mmpcmcou comm ecu mo muo.cum.o ccocuaochc me. new muo.cum.o ccozu:Om mm mco.moL >0 muu.cum.o u.umLooeoo c.cu.3 mo.um.L0uumcmco >ucoau.umcoo mo cOm.cmanuii.~> o.nmh 187 explained by the data summarized here. It will be recalled from chap- ter four that among Democrats the independent variables most directly related to roll call voting variables were percent urban, percent foreign stock, percent with private elementary education, median family income, percent sound homes with all plumbing, median home value, and median rent. Poor education and percent farmers were strongly negatively related to the voting dimensions. The means for these variables for southern districts are all significantly different than those for the nonsouthern districts. These differences consistently go with the signs of the correlations; that is, the variables directly related to liberal voting have higher means for the nonsouthern districts than for the Southern ones. The inversely related variables, percent with poor education and percent farmers, have higher means for the southern districts. Variation in constituency characteristics of Democrats is largely accounted for by the less urbanized character of the southern States. In fact if Table V-2 is compared to Table IV-3 where the two parties are compared in terms of the 21 constituency variables, it can be noted that on thirteen of the twenty-one variables the F statistics of mean difference are larger in the south-nonsouth comparison than for the Republican-Democrat comparison. Much of the variation, then among Democratic districts is accounted for by the southern districts. These conclusions fit with the data of Table V-l. The variation from party unity on the dimensions of voting are also accounted for by the southern representatives. That is, the dimensions which have strong correlates among the constituency variables are those which divide southern and nonsouthern representatives. Referring back to Appendix F 188 for Democrats, 1 took the highest correlation of each dimension. 1 ranked the dimensions from low (absolute value) to high and correlated the ranks to those in Table V-l. rs = .856. This simply means that most of the variation in Democratic voting is variation by southern Democrats from their nonsouthern colleagues; similarly the independent variables which correlate most strongly with the dimensions of voting are those which distinguish southern and nonsouthern districts. The apparent constituency relevance to the voting of Democratic representa- tives is summarized by differences between southern and nonsouthern voting and constituency characteristics. There is no subset of Republican districts which is bound to- gether in a fashion like the southern Democrats. The midwestern and plains states have been known as the heartland of Republican electoral strength, but even there Democrats have made strong inroads in the period since the Depression. Actually Maine and Vermont have been most faithful to the Republicans in the Presidential elections between 1932 and the present (see Jones, 1965), but they are of small importance in the House with only three representatives. Possibly Republican representatives can be separated more meaningfully on the basis of one or more continuous variables. One could divide the 178 Republicans according to whether they are high or low on a particular characteristic such as percent urban, percent Negro, percent foreign stock, percent farmers, etc. There is the general feeling that small town, native, white American Republicanism is a type to be contrasted with a somewhat more heterogeneous, cosmopolitan, and sophisticated Republicanism. In short, the major difference is be- tween liberal and conservative Republicans. Among the constituency 189 characteristics used, there really is no satisfactory variable for making the distinctions described. Percent urban is too broad because the census definition is so inclusive. Percent farmers is a narrow occupational variable which is not necessarily indicative of small town values. Percent foreign stock may pick up too many things to be useful, especially among Republican districts. It does not distinguish between foreign stock which was readily assimilated into the fabric of the communities and those which were not. For example, of the thirty ”most Canadian Districts” of the 88th Congress, fifteen were Republican (CQ Census Analysis: Conqressional Districts of the United States, 1964, p. 1865). The ease with which these individuals have entered the com- munity is greatly different than has been the case with the Irish, southern Europeans or Puerto Ricans. The low variation associated with percent Negro, noted in chapter four, discourages use of this variable. If the midwest has been the center of gravity for the Republican party, its heaviest counterpoise has been the east. Probably most effec- tive in the presidential nucleus, the ”Eastern Establishment” has stood for different brand of Republicanism than the remainder of the Republi- cans have. The professional managers of finance and industry, located particularly in the eastern cities, have preferred the Republican party over the Democratic party as it has opposed the demands of labor and social liberals. However, they have adopted a much more pragmatic view of the role of government in the national economy. Preferring a stable economic growth rate, they have been less agitated by governmental spend- ing and unbalanced budgets (see Monsen and Cannon, 1965, pp. 24-63). Meanwhile, hopeful Republican candidates in these states confront a 190 political opportunity Structure which are mightily affected by the dominating votes of urban and suburban residents. A convenient division of Republican districts results from a combination of the New England and Mid-Atlantic states. By this defini- tion the east includes Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont, Delaware, New Jersey, New York, and Pennsylvania. Actually, the inclusion of Delaware and Rhode Island is of no conse- quence because the three districts they contain were all Democratic in the 88th Congress. Republicans held almost half of the eastern constit- uencies with 54 out of 109. These 54 are less than a third of all the Republican seats in the 88th Congress (Republican total; 178). While it may be argued that congressional voting is not the most obvious place to look for cleavage in the Republican party, this tendency has been noted in the literature. Truman (1959, p- 179) noted that among House Republicans ”there was an evident connection between geographic region and position in the [roll call voting agreement score] structure. With a few exceptions the men who belonged to the blocs on the left [liberal] side . . . were from the east and those in the blocs on the right [conservative] were from the middle west. The cleavage on these votes thus resembled that among the Senate Republicans.” The analysis here will make the comparison more systematic although it means includ- ing a number of outstate, small town, native white constituencies in Pennsylvania and New York, and rural New England districts. Table V-3 establishes the degree of intra-party cleavage among Republicans. The regional differences are not nearly as great as among the Democrats, but most of the differences are considerable. Differences 191 Table V-3.--The degree of differentiation between eastern and noneastern.Republicans on 16 dimensions of roll call voting (Kruskal-Wallis H coefficients) Conservation . . . . . . . . . . , . . . . . . . . . . .36a Foreign trade . . . . . . . . . . . . . . . . . . . .99a Agricultural policy . . . . . . . . . . . . . . . . 2.52a Party loyalty . . . . . . . . . . . . . . . . . . . 4.52 Consumer protection . . . . . . . . . . . . . . . . 7.68 Civil liberties . . . . . . . . . . . . . . . . . . 7.77 Spending . . . . . . . . . . . . . . . . . . . . 12.33 Orientation to debt . . . . . . . . . . . . . . . . 15.29 Negro rights . . . . . . . . . . . . . . . . . . . . 17.05 Aid to education . . . . . . . . . . . . . ... . . . 22.79 Space spending . . . . . . . . . . . . . . . . . . . 30.19 Social welfare . . . . . . . . . . . . . . . . . . . 31.18 Foreign policy . . . . . . . . . . . . . . . . . . . 32.81 Labor vs business . . . . . . . . . . . . . . 45.45 Orientation to professionalism . . . . . . . . . . . 46-85 Urban improvements . . . . . . . . . . . . . . . . . 57.76 acoefficients of difference where eastern Republi- cans have lower scale score ranks than noneastern Republi- cans. greater than 10.837 are significant at the .001 level. Because the dis- tribution of H coefficients is not as broad as among Democrats there is less apparent structure in the distribution and the order of the items is quite different than that for the Democrats (This is evidenced by the low correlation between the two rank orders: r5 = .338, P> .05. The cor- relation is not significant.) There is not as distinct a difference in the cleavage among Republicans by social policy dimensions compared to economic policy dimensions. There is a good deal more unity among Repub- licans than among Democrats on Negro rights, Party loyalty, Civil liber- ties, and Spending. The cleavage among Republicans is nearly as great as among Democrats only on Foreign trade (where differences are not significant within either party), Agricultural policy, Space spending, and Orientation to professionalism. There is not the tendency in this 192 distribution for economic policy coefficients to be clustered at one end while social policy coefficients are cluStered at the other. The differences between eastern and noneastern Republican repre- sentatives are accounted for by more liberal voting on the part of the eastern Republicans (see footnote 1, EEELE) on all dimensions except Conservation, Foreign trade, and Agricultural policy. They clearly suggest that it is in the east that Republicans tend to converge upon the policy positions of the Democratic party. This is reasonable, given the competitive position of the two parties in party identifications (see Campbell, _1 _1., 1960, Table 7-4 on p. 158) in the east and the advantageous access to the national opportunity structure available especially in Massachusetts and New York (Schlesinger, 1966, pp. 27-33). Similarly there are differences between the constituency characteristics of the eastern and noneastern Republican districts on precisely those variables which relate to the dimensions of voting. The difference on percent urban is not great but the differences on percent foreign stock and percent farmers are very sharp. The differences on economic variables are all significant too and given the earlier discussion of how scale Scores are distributed in relation to income, the higher income levels of the east are a basis for more liberal voting. Comparing the F statistics in Table V-4 to Table IV-3, where Republicans and Democrats were contrasted, it can be seen that most of the intra-party differences for Republicans are smaller than those between the parties. This was not the case for Democrats. The greater overall homogeneity of Repub- lican districts is underscored by this fact. 193 .moo. coca mmo. m. .o>o. oucmo.m.cm.m ecu umsu mOumo.oc. 00. oeucoaOL ozhm 00. .0.n~ -.m 00.0. 0m.~ :.m Loecmu.N _o. 3.0 ng 2.0: mm.» 3.9. 29.8: :58 2.; N N0. ~N.m 0..m 0m.m: .0.0 ...m: mcoxcoz cm..ou 00.53 N m.. N0.~ mm.m. nm.~n ~0.m. NN.0N acme cm.ooz 00. 00.m 0..Nm 00.m.. mu.:: 0N.mm. o:.m> 08o: 00.00: :0. m0.: Nu. mm.~ 0m. :0.m u.c0 0c...030 Lea chmLoa 00.00: 00. Nu..m mm. 00.0 mm. mm.m u.c: me...ozu Lon meooL cm.oox 8. to. am... on...“ $0 5.8 9.3236 :a £3 3:5 258 N m.. .0.~ 00.0 -.m0 ::.~. :..N0 mu.c: mc...ozo 00.00000 Loczo N m... :o. 3.. mm... 05. 3.: 86.3.2: N 00. RN... 0m.0m~. m0...0m 0m.~N.. 0~.0n:0 oeouc. cm.ooz 00. N.. 00.. N0... :0. 00... co.uouaeo cm.oox 0.. on.N ~0.m A..0 mm.m N..m co.umu:oo ome..oo 00.3 N m... cm. mu... 9...? 3.0 $23 838%... 39.8 to: £5 N m0. :0. Nm.m 0~.m m:.. m..m co.umo:oo 3o. Lu.) N 00. mm.- N..m 00.m. -.0. .N..~ co.umuaoo >Lmucoeo.o Oum>.ca Lu.) N 00. -.~m :m.~ N..m~ :N.~ :0..m omm cm.ooz 00. mm.00 -.m Nm.0. 0:... 0m.0~ xoOum 00.0Lou N mm. .:. 0N.: 00.m 00.m :m.m ocmoz N N0. om.m 0m..~ :m.m0 00.0w 00..N cont: N N0. 0.. mn.mm 0m.0m ~0.mm mm.m~ omcmsu co.um.:aom N we o.u co.u emu: co.u and: co .o>00 im.umum im.>00 im.>o0 . oucmu m .ocm ocm mo.nm.cm> ucoocoaooc. -.c.cm_m , -ccnom -ecnum memo: coozuom .mcmu..eaao¢ mc~0..0000¢ oucoLowm.0 ..cLOHmmocoz cucummu L n r i mmomeOu £000 050 wo muu.cum.o accommocOc :N. van muu.cum.o ccoummo :m mco_00c >0 muu.cum.o cmu..n:ao¢ c.nu.3 mo.um.couumcmco >ocoau.umcou mo com.Lmanuii.:i> 0.0m» 194 The regional analysis fills out the findings presented in chap- ters three and four. It locates the major sources of variation in voting within each party and Shows the differences in the constituency characteristics of the regions. The constituency differences which occur fit with the voting differences. The links established in chap- ters three and four between percent urban, percent foreign stock, per- cent with low education, median family income, and percent farmers Show up as the variables of the parties. But the differences in the constit- uency variables are much less distinct than the differences in roll call voting, and the differences in voting within the parties, certainly of noteworthy proportions, do not reach the proportions of the conflict between the parties on most dimensions. For example there is relatively high internal disagreement within each party on the urban improvement dimension, yet the conflict between the parties is a good deal sharper than the conflict within each. This, it seems to me, is very impres- sive evidence about the importance of the legislative parties in struc- turing the voting in the House of Representatives. Competitiveness There has been a good deal of lamentation among political theo- rists about the lack of electoral competitiveness in the primaries and general elections of congressmen. The argument is simple enough. If Ford keeps General Motors improving its automobiles, competitive parties ought to result in better representation. Competitive corporations do not necessarily mean that every community contains competitive franchises. Competitive corporations do not mean local sales representatives are com- petitively aggressive. Corporate competition does not assure competitive 195 models or units. Similarly nationally competitive parties are not com- petitive in all, not even the majority of electoral units, whether these units be states, cities, legislative districts, counties, etc. The more difficult task is that of evaluating representation--different theorists would tout different standards of evaluation. Some argue that a competitive situation would assure conscientious service by represen- tatives. It is the Madisonian argument of The Federalist, No.451 that, “Ambltion must be made to counteract ambition,” and ”that the private interest of every individual may be a sentinel over the public rights.” Thus a competitive electoral situation makes the candidates serve their own interest--win elections--by serving and promising to serve in a way pleasing to constituents. Others may argue that a competitive system is needed, but that, particularly in the legislative halls, not all parts ought to be competitive lest a landslide election altogether remove a minority party from office. Another argument is that with wide- spread competition and turnover of office holders there will be insta- bility, great repetition of effort from term to term, and generally a less competent legislative branch. There are many more value implications which need not be re- viewed here. There is the more practical question whether competitive- ness seems related to variations in the behavior of political authori- ties. In chapter one I reviewed several pieces of research relating to this question and concluded that because the findings are not consistent, the question remains an open one. I have a situation in which the ques- tion can be examined. Competition for House seats is a variable, and several indices can be used. Having measured voting, the question can 196 be reduced to considering whether there is any difference between the voting of representatives from competitive districts and those from less competitive ones. My measurement of competition is based upon expert judgment rather than more objective criteria. It is 99's (Congressional Quarterlngeekly Report, November 2, 1962, pp. 2092-2095) competitive- ness rating which was rendered into a nine point Scale. 99 categorizes competition from as follows: Safe Republican Leaning to safe Republican Leaning Republican Doubtful to leaning Republican Doubtful Doubtful to leaning Democratic Leaning to safe Democratic Safe Democratic \D\I0\\n-JTWN—- I Simply coded each district by the accompanying numbers. The rationale for selecting this rating includes a couple of considerations. First, although the rating is subjective it is authoritative. In chapter one I argued the importance of the perceived importance of constituency variables. The same considerations apply here. If competition is to affect behavior, it must be perceived. I could, for example, have con- structed an objective index based upon the relationship between the winners' votes and their nearest rivals' votes, concluding, for example, that a difference of three percent between the winners' and losers' vote is more competitive than a seven percent difference. Schlesinger, (1955), has argued that turnover is an appropriate basis for an index of competition. Both of these kinds of criteria are relevant, yet both imply problems. The latter approach requires looking at several elec- tions, and the length of the period used materially affects the evalua- tion. Reapportionment makes it difficult to use for all districts. 197 The percentage approach involves an equal interval measurement which may not correspond to the way the contest was perceived by those involved in it. Several elections resulting in the same margin of victory may have been differently understood by the participants. In one district there may be a narrow but Stable margin by one party over the other. In another district patterns may be more shifting. Simply in percentage terms there may be no difference in competitiveness, but I would expect that in the latter district both competitors would be running a scared campaign; in the former that would be less likely. An advantage of the ‘99 index is its authoritativeness. Its evaluations are often known by the participants to the contests. They are reported in the news media. I feel quite certain that any candidate whose position is no better than ”leaning” one way or another is worried about the outcome and, if he wins, cannot help but be concerned about what might happen two years later. I constructed three indices. The first, electoral competitive- ness, has been described. The second concerns the other part of the double jeopardy of a candidate, namely, the primary. This is more sub- jective still because 99 does not comparatively rate these contests. However, the 99 Weekly Reports describe the outlook shortly before the primaries. I examined these reports for all the eventual winners. There are three kinds of contests: (1) the winner met little or no opposition; (2) the winner met serious opposition, but was clearly favored to win; (3) the outcome was in doubt. The district was coded according to the eventual winner's primary contest: 198 Republican - little or no contest . Republican - serious opposition, but favored Republican - contest in doubt Democratic - contest in doubt Democratic - serious opposition, but favored Democratic - little or no contest \OQNWN— Every district has a primary competitiveness score. The gap between the parties relates to the third index, the combined competitiveness score. Election is usually the greater hurdle. However, assuming the value of competition, it is better that primary competition occur than that there has been none at all. It remains, nevertheless, a poor sub- stitute for electoral competition (Turner, 1953; Key, 1956). Thus to get a combined index one must establish a rank relationship between the definitions of electoral competition and primary competition. The rank- ing used here is arbitrary, of course, but because incumbency is such a strong advantage in primaries and because the data for evaluating com- petitiveness in these contests are less systemmatic than that for elec- tions, I prefer a cautious classification of primary competitiveness for combination with electoral competitiveness. Therefore a ”doubtful” pri- mary is given the same rating as a ”leaning” election contest. The com- bination competition score for a district is taken as the election or primary score which is closer to 5. A couple of examples make the point clearer. Representative Tuten (D, Ga-8) had a safe seat in the 1962 election. However, his primary opposition was strong and the outcome was in question. His primary competitiveness score is 7, his electoral competitiveness score is 9, and his combined competitiveness score, be- cause the primary test was more serious than the election, is 7. (see C9 Weekly Reports, pp. 1465, 1545, 1546). Representative White (D, 199 Idaho 1) had no primary opposition, but his election contest was doubt- ful to leaning Democratic. His combined competitiveness score is 6, the same as his electoral competiveness score. The statistical technique is analysis of variance as before. Two groups are formed for each index of competition--more competitive and less competitive. Among the Democrats 63 of the 257 representatives (24.5%) won office in districts where the electoral competitiveness ranged from (3) leaning Republican to (7) leaning Democratic. The re- maining 194 won in (8) leaning to safe, or (9) safe Democratic seats. For primary competiveness 38 of the winners (14.8%) came from districts where the primary was (7) in doubt. The analysis using the combined competitiveness scores compares the members who faced no sharp opposi- tion in either primary or election with those whose primary was in doubt or whose seat was no safer than ”leaning Democratic” in the elec- tion. 86 members (33.9%) fit in the latter category. Table V-5 presents the KruskaI-Wallis H coefficients of differ- ence between the voting ranks of the groups on each dimension for each index of competition. Overall the differences appear to be meager. This is especially true in comparison with the differences between the parties and between the regional wings of the parties. It should be kept in mind, however, that at conventional levels of significance relatively small H coefficients give evidence that groups are statisti- cally different from one another. At the conventional .05 level of significance an H coefficient equal to or greater than 3.841 is signifi- cant. The more demanding .01 level requires an H equal to or greater than 6.635 (with one degree of freedom). 200 Table V-5.--The degree of differentiation between Democratic Representa- tives from more and less competitive districts on 16 dimensions of roll call voting (Kruskal-Wallis H coefficients) Electoral Primary Combined Dimensions Competi- Competi- Competi- of Roll Call Votinga tiveness tiveness tiveness Negro rights 22.89 ‘ 3.42b 8.52 Aid to education 7.06 6.94b .16 Consumer protection 6.29 .03 3.65 Conservation 2.73 2.41b I9 Orientation to professionalism 2.15 6.52b .15b Civil liberties 11.23 1.31b 4.45 Social welfare 4.63 3 12b .42 Foreign policy 11.23 2.15b 3.34 Foreign trade 1.31b .14 .13b Urban improvement 18.71 1 52b 8 06 Labor vs business 5.56 1.21b 1.72 Spending .08 1.73b .Olb Agricultural policy .00b .84b .12b Orientation to debt 3.14 2.08b 23 Space spending 4.63 1.96b 1.49 Party loyalty 10.75 3.57b 2.09 8The dimensions are ordered according to the degree of inter- party differentiation established in Table IV-l. bCoefficients of difference where representatives from more com- petitive situations have lower Scale score ranks than their fellow par- tisans. The general impression from the data is that electoral competi- tion is the most meaningful of the three measures of competition. Repre- sentatives from competitive primaries vote significantly differently (P - .05) only on the Orientation to professionalism dimension and the Aid to education dimension. Using the combined competitiveness index significant differences occur on the Urban improvement, Civil liberties, and Negro rights dimensions. Electoral competitiveness reveals more and sharper differences, especially on Urban improvements and Negro rights. 201 More important is the direction of the differences between the groups. The differences in voting by representatives from more and less competitive districts are accounted for by the generally higher scale scores by representatives from competitive districts. Except for the Agriculture and Foreign trade dimensions, where differences were insig- nificant, Democrats from competitive districts vote more liberally than representatives from safer seats. This is not the case for Representa- tives who survived doubtful primaries. The differences, thoUgh smaller, were in the opposite direction, except of Consumer protection and Foreign trade. Representatives from comeptitive primaries (half from the South) are different from the rest by the fact that they have lower scores on the dimensions of voting. Most Democrats with doubtful pri- maries were much safer in the election than the primary. This is par- ticularly true among the southern representatives who make up half the members surviving doubtful primaries. The findings using combined competitiveness are, understandably, muted by the counter trends of the two kinds of competition. rElection competitiveness dominates this divi- sion, so on most dimensions the representatives from the more competi- tive districts tend to have higher scale Scores than those from less competitive districts. (On four dimensions where differences are tiny competitiveness goes with lower scores: Orientation to professionalism, Agriculture, Spending, and Foreign trade.) ‘Among Republicans there is a larger proportion of electorally com- petitive seats than among Democrats. 52 of I78 Republicans (29.2%) won office in contests that ranged from (3) leaning Republican to (7) lean- ing Democratic. The remainder were from safer districts. Only 15 202 districts (8.4%) had doubtful primaries in 1962.1. The combined competi- tiveness index reveals 59 districts (33.1%) were represented by congress- men who survived a doubtful primary or won in an election contest wHiZh‘ was no safer than ”leaning Republican.” Table V-6 shows that competitiveness is much less related to variation in Republican roll call voting than it seems to be for Demo- crats. There are no statistically significant differences between repre- sentatives from electorally more competitive and less competitive dis- tricts on any of the dimensions. Two, Orientation to professionalism and Aid to education, provoke significantly different responses from representatives from doubtful primaries. None of the differences are significant using the combined competitiveness index. In contrast to the Democrats, there are no distinctive directional contrasts in the differences by primary as compared to electoral, competition. For all three indices the representatives from more competitive districts generally have lower scale scores on the voting dimensions than those from safer seats. The exceptions are Agricultural policy, Spending, Conservation and Space spending, and on all but the latter the slightly higher scores are almost immaterial. The findings, stronger among Democrats than among Republicans, tend to confirm findings by Huntington (1950) that members from competi- tive districts Show more marked liberalism and conservatism than do par- tisans of safer districts. What is noticeable in these data is that the differences according to electoral competitiveness tend to be greater on the social policy and foreign policy dimensions where the interparty roll call voting cleavage is relatively lower. 1 think this simply 203 Table V-6.--The degree of differentiation between Republican Representa- tives from more and less competitive districts on 16 dimensions of roll call voting (Kruskal-Wallis H coefficients) m _====== Electoral Primary ‘ Combined Dimensions Competi- Competi- Competi- of Roll Call Votinga , tiveness tiveness tiveness Negro rights 1.82 .18 2.33 Aid to education .30 3.21 .26 Consumer protection 1.42 .OOb 1.16 Conservation .38b 1.05b 1.13b Orientation to professionalism 2.92 2.76 2.87 Civil liberties 3.15 .56 2.29 Social welfare 1.54 .53 1.03 Foreign policy 1.26 .32 .60 Foreign trade 1.12 .l8b .63 Urban improvement .27 1.16 .15 Labor vs business .00 2.68 .03 Spending .32b .41 .33b Agricultural policy .13b .05 .44b Orientation to debt 1.21 .80 1.09 Space spending 2.22b .01 2.63b Party loyalty 1.31 .02b .83 3The dimensions are ordered according to the degree of inter- party differentiation established in Table IV-l. bCoefficients of difference where representatives from more competitive situations have higher scale score ranks than their fellow partisans. indicates that on economic and Agricultural policy dimensions where the two parties have tautly drawn partisan lines, there is not a great deal of variation among the members which is systematically related to competi— tiveness. The dimensions which do not as sharply divide the legislative parties include variation which, for Democrats in competitive districts, stimulates more liberal voting and, for Republicans in competitive dis- tricts, stimulates more conservative voting than among their respective fellow partisans from less competitive districts. Representatives from 204 competitive districts apparently attribute more relevance to inter- party differences than those from less competitive districts. This even shows in their ”purer” positions on the party loyalty dimension. They extend and exacerbate the interparty conflict to dimensions of social policy in a way that representatives from less competitive dis- tricts do not. Ambition In chapter one I raised the question of whether the political ambitions of representatives might differentially affect their roll call voting behavior. Schlesinger (1965 and 1966) has developed most of the notions to be examined here. Ambition is relevant to policy voting be- cause: If anyone is going to search for solutions, it is the man whose career depends on finding solutions. The politician with Static ambitions is far more likely to be driven by immediate pressures, whether it be the pressure of opinion, party, or special interest groups. Only the man with progressive ambitions is driven to explore current policies in the light of future consequences, for his future career is at stake. In his calculations, todays opinions can be discounted, tomorrow's put at a premium. For this reason the man with progressive ambitions is more likely to seek to lead and direct opinion, not necessarily from any idealistic sense of what is right, but from his own need to secure opinion for his future goals (Schlesinger, 1966, p. 209). A careful examination of differences between those with progres- sive ambitions and those without them is difficult for a number of rea- sons. Specifying who has progressive ambitions is hazardous. Ambition certainly is a psychological syndrome. I have merely identified as pro- gressively ambitious those members of the 88th congress who have run for 205 higher office between 1963 and the beginning of 1967.2 Only twenty-one are treated; ten Democrats and eleven Republicans. Most of them sought higher electoral office, usually a U. S. Senate seat, and most were de- feated. It is worth noting that there are two sorts of progressives among these twenty-one. Those with low seniority who hold a precarious position in a competitive district and have everything to gain by taking a chance on winning higher office; for example, Bolton of Ohio and Mc- Intire of Maine. Similar are those whose seats were being apportioned out from under them as was the case for Staebler of Michigan and Sickles of Maryland. Others relinquished safe seats, sometimes with substantial seniority, to contest for higher office; Duncan of Oregon and Bray from Indiana provide examples. The smallness of the number of representatives from the 88th Congress who revealed progressive ambitions is consistent with Schlesinger's (1966, p. 209) characterization of Con- gress, and particularly the House, as an institution which fosters stat- ic ambitions and responds to contemporary opinion. The question remains whether those who try to move on to other offices evidence a willingness to guide opinion rather than merely reflect it in their roll call voting. 2Among the representatives scaled, 10 Democrats and 11 Repbuli- cans sought higher office and were classified as progressively ambitious. The Democrats are: Bass (Tennessee-6) Ran for U. S. Senate in 1964 Duncan (Oregon-4) Ran for U. S. Senate in 1966 Gill (Hawaii-AL) Ran for U. 5. Senate in 1964 Harris (Arkansas-4) Accepted a federal judgeship in 1966 Harding (Idaho-2) Ran for U. S. Senate in 1966 Hemphill (South Carolina-5) Accepted a federal judgeship in 1966 Montoya (New Mexico-AL) Ran for U. 5. Senate in 1964 Roosevelt (California-26) Ran for Mayor of Los Angeles in 1965 Sickles (Maryland-AL) Ran for Governor in 1966 Staebler (Michigan-AL) Ran for Governor in 1964 206 The differences in voting behavior are hypothesized from what Schlesinger (1966, p. 137-139) has described as ”third order organiza- tional tensions."3 Seeking higher office in broader and more varied constituencies, ordinarily they may be expected to respond to their anticipated electoral strength in the broader constituencies. Therefore progressives might be expected to vote in a more liberal fashion than their statically ambitious colleagues. Democrats need to be responsive to the urban majorities and their demands for progressive social and economic policies. Ambitious Republicans Similarly would be-expected to be more liberal than their party generally because of their need to Two other Democrats, Shelley (California-5) and Thornberry (Texas-10), took higher offices. However, their replacements were scaled on more dimensions, so the voting records of Shelley and Thornberry are not part of the data treated. Progressively ambitious Republicans are: Avery (Kansas-2) Ran for Governor in 1964 Bolton (Ohio-11) Ran for Congressman at large in 1964 Bray (Indiana-7) Ran for Gubernatorial nomination in 1964 Bruce (Indiana-ll) Ran for U. S. Senate nomination in 1964 Ellsworth (Kansas-3) Ran for U. S. Senate nomination in 1966 Griffin (Michigan-9) Appointed to u. 5. Senate in 1966 Hoffman (Illinois-l4) Ran for Illinois Secretary of State in 1964 Lindsay (New York-17) Ran for Mayor of New York in 1965 Lloyd (Utah-2) Ran for U. 5. Senate nomination in 1964 McIntire (Maine-2) Ran for U. S. Senate in 1964 Taft (Ohio-AL) Ran for U. S. Senate in 1964 William E. Miller (New York-40), the Republican candidate for Vice President in 1964, is not included here because he had previously gone on record as retiring from the House at the end of the 88th Congress. He is treated as having discrete ambitions. 3I have not tried to identify those with ambitions for legislative leadership positions. These bring about what Schlesinger (1966, pp. 135-137) describes as ”second order organizational tensions.” I do not think this interferes with the analysis of this paper. Such progres- sives may be expected to be less liberal than those who seek higher office outside the House. ALegislative leaders tend, as MacRae (1958, pp. 289-296) and Truman (1959, pp. 193-236) have reported, to be in the middle of their parties in terms of the liberal-conservative distribu- tions of their parties' roll call voting. 207 converge on the proposals of the opposition party, which is likely to be stronger in the larger constituency than the smaller one. Republicans, likely to be the minority party in most large constituencies, can win partly by not arousing the antagonism of low income, less educated, working class, Democratic identifying voters. Convergence by being more liberal than the typical sitting Republican representatives may be the way to higher office for Republicans. The data reveal only small differences between the progressively ambitious representatives and the remainder of the representatives. Each party is examined separately and the Kruskal-Wallis H coefficients are presented in Table V-7. Progressively ambitious Democrats are signi- ficantly (P‘- .05) different only on Labor versus business dimension and progressively ambitious Republicans are significantly different only on Civil liberties. Nevertheless there are suggestive trends in the data. Averaging the ranks for each group among the Democrats shows that, al- though the differences are slight, the progressively ambitious group has higher ranks on every dimension. Voting is, as the theory suggests, a trifle more liberal. ~Among Republicans the findings Similarly lend some credence to the hypotheses. The progressives have higher ranks on all the dimensions except Orientation to debt, Labor versus business, Urban improvement, Spending, Party loyalty, and Agriculture. This is consistent with what might be expected. Progressively ambitious Repub- licans Show a more liberal roll call voting pattern on most of the social policy dimensions, Orientation to professionalism, and Foreign policy. They demonstrate their partisan orthodoxy with lower average ranks on economic policy dimensions and Party loyalty. The only 208 dimensions on which pregressives stray from this description is Urban improvement. I would expect more liberal response on this dimension by the progressively ambitious. Instead their ranks are lower. Table V-7.--The impact of progressive ambitions on roll call voting (Kruskal-Wallis H coefficients for each party separately) Dimensions of Roll Call Votinga Democrats Republicans Negro rights .38 1.57 Aid to education .64 .00 Consumer protection .42 1.78 Conservation .08 .05 Orientation to professionalism .20 1.58 Civil liberties 2.45 3.92 Social welfare 2.37 .79 Foreign policy 1.09 2.47 Foreign trade 2.96 .02 Urban improvement 1.29 1.18b Labor vs business 4.17 1-38b Spending .17 .64b Agriculture .44 .00b Orientation to debt 2.18 2.07b Space spending .49 .66 Party loyalty .28 .52b aThe dimensions are ordered according to the degree of inter- party differentiation established in Table IV-l. bCoefficients of difference where representatives with pro- gressive ambitions have lower scale score ranks than their fellow parti- sans. Reelection Basic to the whole notion of a responsive and responsible govern- ing regime are the arrangements making the members of the electorate the judges of their rulers. The votes of the people determine who will be retained in office and who will be denied from holding office. Assuming each congressional constituency is peopled by an informed and interested 209 electorate, it might be hypothesized that those representatives who win reelection are better, more responsive congressmen than those who fail to be reelected. In most constituencies the assumptions are not met; the elec- torate is frequently not interested and generally is uninformed about the issues and how the representative votes in the electorate's behalf (Miller and Stokes, 1963). Second, it is difficult to specify what constitutes ”better, more responsive” behavior for a congressman. The basic datum for predicting congressional district election outcomes is the division of party identifiers in each district. Then there are candidate and issue orientations involving both congressional candi- dates and the other candidates for governor, senator, or president who lead the ticket in the same general elections. In sum, the fact of election or defeat may be far removed from any estimation of how the congressman has or will represent his constituency. The data and techniques of analysis which have been used pre- viously can be applied here to shed some light upon the meaning of reelection. In simplest terms the question is, do reelected represen- tatives vote any differently than those who are not reelected. Given the finding that the major variable affecting voting is party, the question is best examined within each party separately. Given measures of voting behavior during the 88th Congress, the representatives of each party were divided into two groups; those who won reelection in 1964 and those who did not. It is true that the latter group is diverse. It includes those who did not contest along with primary and election losers. Also the reelection test for veterans of the 88th Congress came in the context of the Johnson landslide. 210 Among the Democrats, 223 of the 257 members of the 88th Congress (86.8%) were reelected. Only five were defeated in the 1964 election, eight were eliminated in primaries, and twenty-one did not contest for office. There were no statistically significant differences between those who were reelected and those who were not in their voting on the sixteen policy dimensions. As Table V-8 shows, the Kruskal-Wallis H coefficients ranged from .02 on Conservation to 3.47 on Spending. Look- ing at the voting patterns there is no regular tendency toward liberal or conservative voting by the reelected representatives. On the aver- age they voted more liberally on nine dimensions, but the differences are very slight and there are no regular patterns among either the social policy dimensions or the economic dimensions. There are no apparent ways in which as a group those Democrats who were reelected distinguished themselves from those who were not reelected in terms of role call voting behavior. This is not to say that roll call voting patterns are irrelevant to the reelection of individual representatives; rather, the variation in roll call voting that does occur does not systematically vary with patterns of reelection. Part of the explana- tion of course is that the South is somewhat overrepresented in the loss of incumbents because primary turnover in the South was higher than the party's election losses to Republicans in the 1964 national elections. The findings among Republicans are more substantial. Their con- gressional ranks were thinned considerably in the 1964 elections. Only two-thirds (120 out of 178) of the 88th Congress Republicans returned to office for the 89th Congress. There were 19 who did not contest, including several of the progressively ambitious Republicans; a proportion Table V-8.--Comparison of reelected and nonreelected representatives on 16 dimensions of roll call voting (Kruskal-Wallis H coefficients for each party separately) Dimensions of Roll Call Votinga Democrats Republicans Negro rights 3.44b 3.84 Aid to education 3.38b 2.05 Consumer protection .03 1.01 Conservation .02 4.07 Orientation to professionalism .24 1.85 Civil liberties 1.72b 3.39 Social welfare .62b 7.05 Foreign policy .11b 9.04 Foreign trade 1.09 9.22 Urban improvement 1.53b -23 Labor vs business .02 .00 Spending 3.47 .67 Agriculture .75 .03b Orientation to debt .02b .36 Space spending .14b .33 Party loyalty 2.40b .46 aThe dimensions are ordered according to the degree of inter- party differentiation established in Table IV-l. bCoefficients of difference where reelected representatives have lower scale score ranks than their fellow partisans. similar to that for Democrats that year (10.7% compared to 8.2% for the Democrats). No Republican incumbents lost primary contests. were defeated in the November elections. But 39 The defeats were rather evenly distributed across the regions; 22.2 percent of the eastern Republicans were defeated, 20.5 percent in the north central states,[-1 and 23.9 per- cent of the Republican incumbents in the rest of the country. “The Survey Research Center designation is used here, combining the east north central states of Illinois, Indiana, Michigan, Ohio, and Wisconsin with the west north central states of Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota. 212 Table V-8 shows that those who were reelected had voted signifi- cantly differently (Pia .05) than their colleagues on five policy dimen- sions: Negro rights, Conservation, Social welfare, Foreign policy, and Foreign trade. The differences are slightest, however, on the dimen- sions where interparty cleavage is most sharp. With the exception of voting on Agricultural policy, reelected Republicans had somewhat more liberal voting records than their colleagues who did not return to the House in 1965. This is perhaps an ironic finding given the policy predilections of Senator Goldwater's most ardent supporters. This set of tendencies, slight as they are, support the notion that the minor- ity party needs to converge upon the policy positions of the majority party in order to survive electoral competition. Summary .The findings of this chapter give further confirmation to the inference that the legislative parties dominate the roll call voting patterns in the House of Representatives. The well advertised South-- nonsouth split in the Democratic party is strongly evidenced in the roll call voting data, but party identifications do more than merely _paper over internal cleavages. On only two dimensions is there more conflict within the Democratic party than there is between the two par- ties. The first, of course, is Negro rights and the other is Aid to education. Among Republicans there is one such dimension, again Negro rights, but the internal division among eastern and noneastern Repub- licans is of mild proportions even of this dimension. Other political variables point up only small distinctions in roll call voting. The data do not indicate that representatives from 213 competitive districts tend to converge on the policy positions of the opposition party, rather they seem to impose a more partisan coloration to those policy areas where the parties as such are not squarely opposed. Ambition relates to a somewhat different configuration. Ambitious Demo- crats take a generally more liberal Stance than others of their party. Ambitious Republicans tend to be more liberal on Foreign policy and social policy dimensions in general--they converge on the majority party's positions at least slightly; but this is not true on the most partisanly related dimensions. Where partisan considerations override, ambitious Republicans preserve their partisan standing by voting even more conservatively than their less ambitious colleagues. Similarly Republican convergence is seen in the voting behavior of the reelected representatives when they are contrasted with those who did not return to the House after the 1964 election. The fact of reelection seems to have less significance among the Democrats. The best evidence of convergent behavior in either of the parties or politically distinguishable parts thereof is in the voting of eastern Republicans. Divergent voting behavior occurs with distinct competitive- ness. Yet in a competitive region where the constituency characteris- tics most associated with liberal voting are concentrated and party (organizational tensions are encouraging, there is evidence of convergent behavior even in the voting of the congressmen. The Democratic South, on the other hand, has not converged on the minority party so much as it has failed to Shed its traditional Democratic coloration which has been maintained to assure the superiority of the white society. As Key (1949, especially pp. 665-675) has pointed out the backbone of Southern politi- cal power has been the black belt counties, whose political roots go 214 back to pre-Civil War Whiggism. Key's description of the South and its congressional solidarity made in the late 1940's fits the behavior of the 1960's. In presidential voting at moments of strain the roots of southern solidarity are defined by the split in the popular vote. In the South's representation in Congress, however, no such split occurs, at least on the race issue. The solidarity of the southern spokes- men in congress reflects in part the success of the black belts in converting the entire South to their will and in part a regional aversion to external interference on any question. Yet it is only on the race question that the South presents a solid front in Congress. In the Senate to a relatively greater extent than in the House the factional differences within the Democratic party of each southern State are projected into the voting on nonracial matters. In the House, however, it appears that one of the serious consequences of the one-party system is a much higher solidarity on all kinds of questions than in the Senate. Yet in both House and Senate it is the race issue that evokes the highest degree of southern solidarity. This phenomenon contributes additional sup- port to the proposition that the Negro gives the South its pe- culiar political characteristics.(Key, 1949, p. 667). The price of this solidarity has been very high. In part it has been paid by able southern political leaders who have been denied fulfill- ment of their progressive ambitions at the presidential level. One of the reasons for the strain noted by Key concerning southern solidarity in presidential elections is the fact that there are net presidential hopefuls among the southern political leaders who would want and need to bind state and local party organizations to the party's presidential nucleus. Schlesinger (1965, p. 769) points out that one would think a governor of New York peculiar if he does act not like a candidate for the presidency, but one would consider a governor of Mississippi a fool if he does. The absence of these organizational strands is of incal- culable impact, but is certainly of substantial effect in preventing real unity in the organization of the presidential nucleus of the Democratic party. CHAPTER VI SUMMARY AND CONCLUSIONS In this study I have attempted to explain roll call voting patterns of congressmen in the House of Representatives of the 88th Congress in terms of (a) the social, economic and demographic character- istics of their districts and, (b) the potentially intervening political considerations. These efforts remain to be summarized, criticized, and put into the context of suggested future research. The findings in chapters three, four, and five are based upon the measurements of voting presented in chapter two. While there are some unfortunate aspects of this measurement, discussed below, the structure in the roll call voting evidenced by the scale analyses is surprisingly strong. Guttman scaling is a relatively demanding tech- nique for establishing dimensionality. I imposed even more rigorous criteria than are conventionally used before including items in the scales. The resulting scales have been shown to be strongly unidimen- sional by several standards and criteria (refer to Table II-2). The sixteen dimensions constructed include 138 items, more than half (59.5%) of the roll calls published in the Conqressional Quarterly for the 88th Congress. In elementary terms, each scale provides a technique measur- ing representatives' responses to the question, how far should govern- ment activity extend into each specific policy area; what Should the 215 216 spending level be, how far ought Negro rights be extended, how generous ought assistance to foreign countries be, what should conservation policy protect, etc. Several of the policy dimensions take on a recog- nizable liberal-conservative distribution, but while some are strongly intercorrelated, each is unique and contributes a somewhat different assessment of the broader liberal-conservative designation. Summary of Findings The scaling process was carried out in such a manner that every representative could be placed on each of the policy dimensions. This contrasts with the design of other studies which have separately measured the voting of Republicans and Democrats. Implicit in this design was an attempt to demonstrate that party voting is really only a manifestation of constituency influence. Second, whether party voting could be ex- plained by constituency data or not, the voting measurement would be such that the extent of party cleavage on each of the voting dimensions could be demonstrated. These aspects of the design were very well fitted for the analy- sis. Twenty-one independent variables for each constituency were cor- related to the voting of each representative on the sixteen dimensions of voting. 1 used rank correlations because the scale Scores are Simply rankings, and not equal interval measurements. On the authority of Flinn's (1964) remark that correlation ”coefficients above .4 are re- markably good in a system as complex as the legislative system,“ I defined relationships above that value as ”strong,” those between t .400 and t .250 as ”moderate,” and those less than t .250 as weak. I hypothe- sized a complete description of relationships between the independent 217 variables and the scale scores of voting, and I reported the complete description of findings for comparison to the hypothesized relation- ships. The correspondence between hypothesized relationships and those observed was very low. Though not as hypothesized, there are noteworthy relationships between constituency variables and some dimensions of roll call voting. Several constituency variables relate in a consistent direction with the roll call voting scores; that is to say, are positively or negatively related to liberal voting. Those especially positively related were low education, percent urban, percent Negro, percent foreign stock, and per- cent unemployed. Negatively related variables are especially the per- cent of population change, percent with high school education, percent of owner occupied units, and percent farmers. Several variables which I expected to be important correlates of voting dimensions were weakly and irregularly related to the dimensions of roll call voting. These included the percent with college education, the economic variables (median family income, percent of sound homes with all plumbing, median home value, and median rent), and the percent of white collar employment. Statistical controls to hold constant the effects of percent urban and percent foreign stock reveal that the economic variables are mostly negatively related to liberal voting. Further analysis of median income showed that there are nonmonotonic patterns between it and some of the dimensions; that is, low income districts are represented by liberal voting representatives, but representatives from high income districts do not vote as conservatively as those from medium income districts. The highest correlation coefficients occurring in the matrix of relationships were with the Negro rights voting dimension and its 218 highest positive correlate was percent foreign stock. However all variables except percent Negro, percent with low education, median number of individuals per dwelling unit, and percent farmers are posi- tively related to liberal voting on Negro rights. This dimension is also the one on which there is the least party conflict. Actually the distribution of scale scores on this dimension indicates Strong support by northern Democrats generally, moderate support by Republicans tend- ing to be strong among eastern representatives, and almost no support by southern representatives whether Democrats or Republicans. For purposes of comparison I grouped together six dimensions of social policy (Civil liberties, Urban improvement, Social welfare, Con- sumer protection, Negro rights and Aid to education) and four dimen- sions of economic policy (Spending, Labor vs. business, Orientation to debt, and Space spending). In general, correlations with social policy dimensions are stronger than those with the economic policy dimensions. Surprisingly this is true for each of the economic variables; median in- come, percent of homes with sound plumbing, median home value, and median rent. Included among the variables more strongly related to economic policies than social policies are three education variables (low education, percent with high school education, and median education) along with percent of owner occupied units and median rooms per unit. Voting on agricultural policy is not related to an urban-rural division of constituencies. Foreign policy voting is most positively related to percent urban and percent foreign stock, and most negatively with percent farmers. ‘Party loyalty voting is most related to the per- cent of the population with a low education. For the rest of the 219 dimensions of voting the correlations tended to be moderate to weak, consistently lower than I had expected. Given the modest level of the correlations between the constit- uency variables and voting behavior, I looked at the correlations be- tween party and voting. Except on the Negro rights and Aid to education dimensions, which aroused the least party cleavage, the correlation of party to voting was higher than any of the relationships between con- stituency variables and voting scale scores. On these fourteen dimen- sions the correlations range from .490 to .821. Half of these are .700 or higher. The dominating effect of party is too obvious to avoid. Therefore the rest of the analysis deals for the most part with dif- ferences between the two parties and the relationships of constituency variables to intraparty variation and cleavage in voting. The two parties are divided most sharply on the economic policy dimensions and, of course, party loyalty, but the divisions on every dimension other than Negro rights are statistically significant. There are statistically significant differences between the constituencies of the two parties on most of the independent variables, but the keenest differences are on measures of education, percent Negro, and percent of owner occupancy. There are more regular and stronger correlations of constituency characteristics with roll call voting scores among Democrats than among Republicans. The strongest correlates continue to be percent urban, percent of foreign stock, and percent farmers. .Percent owner occupancy, which correlated with several dimensions before separation by party, tends to explain little of the within-party variation. The economic 220 variables Show stronger relationships with voting among Democrats than among Republicans, and for the most part the relationships are positive; that is, the higher the median income in Democratic districts, the higher the support for Civil liberties, Urban improvement, Party loyalty, etc. The data in chapter five suggest that the stronger relationships between constituency variables and voting within the Democratic party are substantially explained by the intra-party cleavage between the southern and nonsouthern members of the party. The constituency vari- ables most highly related to liberal voting such as low education, all the economic variables, percent foreign stock, percent farmers, as well as percent Negroes are those which show significant differences between southern and nonsouthern districts. There are, of course, substantial differences in the voting of the southern and nonsouthern wings of the party, but the conflict tends to be highest on social policy dimensions, Foreign policy, and Party loyalty. Conflict on the economic policy dimension is a good deal lower, and it is lowest of all on Agricultural policy and Foreign trade. The generally lower variance in roll call voting among Republicans is accompanied by less clearly defined regional wings in the party. There are noteworthy and statistically significant differences between the eastern and noneastern members of the House, but they are not nearly as sharp as those among the Democrats. In general eastern Republicans vote more liberally than their noneastern colleagues and the differences are greatest on Urban improvement, Orientation to professionalism, and Labor vs. business. The direction and character of these voting patterns by eastern Republicans support the notion that in 221 a competitive situation with access to the national opportunity struc- ture available, members of the minority party tend to converge upon the policy positions of the majority party. There are constituency differences that support the intra-party voting cleavage, especially on percent of farmers and percent of foreign Stock, but again there are not the sharp differences in constituency characteristics between the regional wings of the Republican party that exist in the regional wings of the Democratic party. Examining the relationship of district competitiveness to voting in the House, I found confirmation for Huntington's (1950) findings that Democrats from competitive districts are more liberal in their voting than their fellow partisans from less competitive districts, while Republicans from competitive districts are more conservative than other Republicans. The differences are clearer among Democrats than Republicans. Primary competitiveness is less related to voting dif- ferences than electoral competitiveness, and combining indices of each does not shed further light upon the differences. I hypothesized more liberal voting for congressmen with ambitions for higher political office. The differences were slight but generally in conformity with the expectations. Ambitious Democrats, members of the 88th Congress who have Since run for higher office, voted more liberally on every dimension. On the dimension which most Sharply divides the two parties, ambitious Republicans tend to vote somewhat more conser- vatively than their less ambitious colleagues, but on the rest they vote more liberally. This tendency toward liberal voting by ambitious mem- bers of both parties is most manifest on Foreign policy and the social policy dimensions. 222 Finally I examined whether reelected representatives had voted any differently than those who did not return to office in the following Congress. I expected no systematic differences, and none were evident among the Democrats. However, among Republicans there was a marked ten- dency for reelected members to vote more liberally than those who left offices, particularly on social policy dimensions (with the exception of Urban improvement), Foreign policy and Foreign trade. Republican con- gressmen who survived the Goldwater defeat had shown more convergent roll call voting patterns than those who had been defeated or had not intended to stand for reelection. If there is one aspect of the findings which ought to be under- scored, it is great regularity which partisanship imposes upon roll call voting. 1 think that scholars past and present underestimate the ex- tent of this regularity. They refer to weak resources that party lead- ers have for influencing members' voting and the strong constituency ties. (Dahl, 1967, p. 131; Froman and Ripley, 1965). I think that they are overly impressed with the absence of formal organizational trappings of the legislative parties. However, none of the constituency character- istics examined here comes close to correlating with the scores on the scales of voting as strongly as party does. The deepest regional cleav- ages of each party, substantial as they are, are more ambiguous than the divisions between the parties. Only on a couple of policy dimensions, particularly Negro rights, does this generalization fail to hold. Con- cerning the original focus of this study, constituency characteristics, it should be said that interparty voting differences are accompanied by constituency differences and this is also true for regional cleavages. 223 But these constituency differences are not as sharp nor are they clearly linked with the patterns of policy voting as partisanship is. Why this is the case, goes beyond the scope of this analysis, but I think the findings presented here point out that this aspect of representation should receive more study. It must be studied at the recruitment level to know the relevance of the local party organizations to the repre- sentatives' career opportunities. It must be studied at the legisla- tive level not only in the activities of formal leaders and committee chairmen, but also in the perceptions and activities of those cast in follower roles. These are substantial research tasks, but I think my findings indicate they will be fruitful. Because of the nature of this study I can only conclude that party is at least a catalyst, if not the cause, which brings out much of the unity observed in roll call voting. Apologies and Caveats Often the writer of a research report can be the best critic of his research. It is true, of course, that he may be as blind to certain .other approaches to the problem of the Study at the conclusion as he was when he first considered it, but his closeness to the data, the reporting, and interpreting ordinarily cannot be approximated by those who review that research. He knows at which points he was skeptical of his measur- ing tools, what assumptions he dared not reexamine, where the data gave only the thinnest support to his inferences, and when he was most uncom- ' fortable in the writing process. It is well therefore that a pgst hoc review be made of this study. The notion that constituency pressure affects the behavior of representatives is, as Flinn (l964) remarks, ”a tough weed.” It is much 224 more frequently discussed than tested. There have been serious attempts to test the theory (see chapter one) but the conclusions of these studies, similar to my own, tend to be murky, self-conscious, and tenta- tive. Froman (1963 and 1964) is probably the most confident about his inferences, but others, such as Flinn (1964) and MacRae (1964), who have pursued the same problem, are his sternest critics. I think that the major difficulty is the great gulf between what I measured as the independent variables and the behavior variables which I wanted to explain. The constituency influence theory has sound theo- retical underpinnings. There are compelling normative and empirical reasons for studying the relationships between constituency and repre- sentative. But in the operationalization stage the scholar confronts the overwhelming task of measuring relevant aspects of the constituency. If one has focused on an entire legislative body, this entails studying somewhere in the neighborhood of 25 to 450 constituencies depending upon which legislative body has been chosen. Rarely is there any conveniently available data on all the constituencies of the legislature chosen. Thus, when such data become available, their relevance is more easily assumed than evaluated. Further, the constituency theory has not been elaborated in a way that it can be tested with such data. Aggregate con- stituency data are simply enumerations of certain kinds of characteris- tics as they exist at time X. One must assume a set of implications about being Negro, unemployed, poor, white collar, etc; assume further how these implications are accentuated as these characteristics take proportion in a constituency; and suppose that they are understood in a fairly consistent manner as inputs by all the representatives in the 225 legislative body. The data for the independent variables are in many cases I'central tendencies”--median income, median education, median age. Perhaps the degree of variance around the central tendency is in some cases more important than the median. These data impose a feeling for linear relationships, yet it is likely that thresholds are important. The percentage of blue collar workers is practically unrelated to the dimensions of voting, but at some point the proportion may become sali- ent, yet undetected, in the correlations. Then there is the multiple branched assumption that constituency interests, more or less identifi- able in the characteristics measured, make themselves felt and/or are anticipated and heeded by representatives. Thus I look ruefully back at the hypothesized relationships of each constituency variable with each dimension (see Table 11-3) and the unanticipated findings reported in Table 111-2, but they are not really shocking. The predictions may have been wrong but the fact that they were wrong does not disconfirm the theory because the theory did not specify the hypothesized rela- tionships. For example, my specific expectations concerning the rela- tionships of college education to policy voting were wrong, but this is because I misinterpreted the implications of college education, not be- cause constituents do not affect the behavior of their representatives. It is exactly for this reason that this study can only have standing in the literature as another ”exploratory study.” Other difficulties in the study involve the measurement of roll call voting. I have referred to them several times already. First is the inability to finely distinguish between the members of the House on several dimensions. Obviously with many members and only a few scale points there will be tie scores. But the ties in the scales 1 have 226 presented are inordinately at one end or the other of the distribu- tions. On Foreign policy almost 36 percent of the representatives have a score of fifteen. 0n Negro rights over 38 percent have the highest score of five. Labor versus business similarly has almost 42 percent tied at the highest score of eight. On Consumer protection nearly 33 percent have zero, the lowest score. These measurement considerations have an unfortunate impact upon the basic tool of analysis-rank cor- relation. The necessary correction for ties in the ranks distorts the indices of relationship. While the tie correction reduces the correla- tion coefficient from what it would be before correction, one still cannot be sure whether or not the corrected values are valid measures of the relationship. The political implications of the great number of ties at the extremes of the scales is itself an interesting point. My interpretation is that in the 88th Congress it was unusual for an ex- treme proposal, particularly an extremely liberal proposal, to come to the floor for a record vote. Such proposals either did not emerge on to the floor, or were defeated by voice votes and thus do not appear in the roll call voting record. As I argued previously, it is more likely that they are killed in the committees. If this is true, I suspect that a scale analysis of the 89th Congress would result in even more skewed scale distributions. The reinforcement of liberal elements of the House following the 1964 elections probably is more evident in floor voting than in the committees. Another aspect of the measurement difficulties is in the assump- tions of Guttman scaling. I have belabored the point that it produces ordinal rather than interval level measurement. This has allowed use 227 of only nonparametric techniques of statistical analysis. It may well be that this is an unnecessarily cautious approach. Some practition- ers argue, but not in published sources, that with a large number of respondents structure will appear in a matrix of product-moment cor- relations which would not be at substantial variance from that in a matrix of rho or tau correlations. The former are much easier to de- 1 rive and use. However, this necessitates making the equal interval assumption about the scale scores. Future Research .Population characteristics may well be too superficial a set of data to use as the key for opening up the relationships between con- stituencies and representatives. Much more must be learned about the representatives on the one hand and constituencies on the other. Ex- plaining why a representative behaves as he does requires a good deal more data from representatives themselves--what are their perceptions about themselves, their identification with their party, their constit- uencies, their careers, the issues of the day, the consequences of their behavior, etc. Investigating these and further questions is very dif- ficult because it is often psychologically and politically threatening to those in office. Access to congressmen for depth interviewing is increasingly limited. Unfortunately there are fewer advantages than disadvantages for the congressman who is open and responsive to schol- arly inquiries. I suspect that political scientists are going to have IThe computer center has, with some anguish, allowed me to ex- pend roughly twenty-three hours of CDC 3600 time doing the rho, tau, and partial correlations reported here. The product-moment correlations could have been produced in a matter of minutes. 228 to be more respectful of these primary sources of data in order that they do not become interview hardened and the access available does not dry up completely. There are suggestions of these problems even among state legislators, especially those near research oriented universities. Perhaps more can be done to study constituencies. The powerful imprint of party on congressional voting suggests that there ought to be more study of legislative parties and local party ties of representa- tives. The representative's lines of communication, doubtless latent much of the time, may be laid bare by knowing what elites have access to him and how these elites participate in the recruitment process. Snowiss (1966) suggests some considerations which may be worth pursuing on a much more broadly comparative scale. I think a good many political distinctions may relate strongly to the representational patterns only partly evident in roll call voting analysis reported here. There is another approach to the data I have which probably should be used. That iS factor analysis. Because I wanted to be able to detect and distinguish relationships between particular independent variables and particular dimensions I chose the methods which have been employed. The patterns in both constituency data and the roll call voting dimensions indicate strong intercorrelation. Factor analysis is probably a better technique for assembling the related elements of the data. Like the correlations, these results may locate patterns for fur- ther research into elites, groups, and interest articulation. l have no idea how often scholars feel a sense of disappointment about the evidence they muster in behalf of the theories they examine. I expected clear contours in the relationships between the independent 229 variables and the scales of voting which would render obvious the foci for future research. These expectations were not met in the findings presented here. I suppose this is one of the best rationalizations one can make for the exploratory study. It reveals which lines of approach hold promise and which do not. I have discovered a good deal about patterns of voting in the House of Representatives, but I have not been able to explain those patterns in terms of aggregate district charac- teristics. I believe that more will be revealed about voting patterns as political analysts map the lines of communication, organization, and influence in the constituencies. APPENDICES 230 APPENDIX A The roll calls in each category are identified by the year in which they were voted and the number assigned them by the Conqressiongl nggterly. The underscored items are those which were reflected in the scaling pro- cess . A-l Foreign policy (20 items) 1963: 10, _2_1, 9.1.. 62,95, 100, 112 1964: 11, 13, 15, 119, 51, 52, 60, 61, 62, 16, 86, 92, 110 A-2 Foreign trade (4 items) 1963: 90 1964: 96, 195, 106 B-l Consumer protection (4 items) 1963: .89 1964: 1.11. 46, 91 B-2‘ Conservation (7 items) 1963: 18, 92 I964: 68, 73. 74, 75, 109 C-1 Agricultural policy: farmers versus nonfarmers (1 item) 1963: 4 1964: C-2 .Agricultural policy: liberal versus conservative (9 items) 1963: 19. 17, 3.1.. 34. 84.. 9.8. 99 1964: 44, 93 231 232 Urban improvement (11 items) 1963: 32, 40, 47, 191, 104 1964: 3, 4_, 5, 5.13.. 59, 102 Negro rights (6 items) 1963: 11. 72. 29 1964: 9. 63, 64 Civil liberties (18 items) 1963: .5..é. 7, 8. 22, 23, 29,,39, 42, 43, 59, 105, 111 I964: 535 éé..§&..199..191 Aid to education (6 items) 1963: 53. 60, 85 1964: .111, 112, 113 Orientation to professionalism (15 items) 1963: .19. 15. 59. 70. 80 1964: .19, 17, 18, 19, 20, 21, 22, 53, 67, 72 rEconomic policy [a classification follows] (55 items) 1963: ,9, ll, 12, 13, 24, 27, 28, 35, 36, 37, 41, 55, 45, 4g, 49: 52. 56. 51, 58. 6 , 64, 67. 68, 69. 73.15. 75. 76. 77..12. 33. 87..§§. 39. 93. 102 1964: 6, 7, 12, 26, 27. 33. 35. 45. 54. 55. 56. 57. zz.§z. 90. 94, 103, 107, 108 Social welfare (17 items) 1963: 50. 55. 65. 78. lgé. 109. 115 1964: 29, 3O, 69, 7O, 71, 19, 80, 82, 89, 99 Party loyalty (14 items) 1963: 1, 193, 106, 118, 119 1964: 8, 16.28. 32, 33. 81.. 8.8.. .21.. 97 233 Congress versus the executive (1 item) 1963: 110 House versus Senate (1 item) 1963: 20 House leaders versus dissidents (7 items) 1963: 81, 91 I964: 36. 39. 59. 52. 85 Unique (7 items) 19632 26, 30, 51, 82, 97 I964: 23. 98 Ambiguous or multidimensional (22 items) 1963: 2, 3, 19, 25, 38, 54, 71, 94, 107, 113, 114, 116, 117 1964: 1, 2, 24, 31, 38, 41, 42, 95, 104 Unanimous votes (7 items) 1963: 66 1964: 25. 37, 48, 65. 76, 83 Reclassification of economic policy items 1-1 1-2 Labor versus business (15 items) 1963: 12. 31. 36. 37. 67. 6.8. 69.811. 1964! 6. 7. 12..5£r 57..ZZa 94 Spending (24 items) 1963: 9, II, 13, 24, 35. _44_. 45. £18. 92. 9.3. 73. .73. 75. 77. 93, 102 1964: 19, 35, 45, 54, 90, 103, 107, 108 234 I-3 Space spending: lunar landing versus military objectives (4 items) 1963: 52, 64 1964: 26, 87 1-4 Orientation to debt: liberal versus conservative (9 items) 19631 27. 28, 56, 57. S8, 87, 88, 89 1964: 56 Five items have been excluded from the general economic policy category (55 items). Three would have scaled acceptably, but each had only one dissentor and thus would have contributed practically no discriminatory power to the scales. On two items (1963: 41, 83) Curtis (R Missouri 2) alone opposed more spending. On one labor versus business vote (1963: 79) only Secrest (0 Ohio 15) took a pro-labor position. The other two items were spending items (1963: 76; 1964: 27) with excessive absen- teeism (44.44 and 38.37 percent respectively). Two items (1963: 31, 84) concerning Mexican farm labor were added to the labor items. APPENDIX 8 Content of the items included in each scale. Roll call numbers are from 99. The items are ordered as they have been ordered in the scales; the most liberal item is listed first. On the scale the most liberal item is always on the extreme left of the scale. The marginal totals reported (% affirmative) are those used in this study, as defined in chapter two. A-l Foreign policy (15 items) CR = .950; CS - .880 1964: 110 45.48% affirmative H Res 892. A rule for consideration of H J Res 1183, con- tinuing appropriations for fiscal 1965 until October 10 for those agencies whose regular fiscal 1965 appropriation bill had not been enacted. Rejected 159-193. 1963:61 (Reflected) 45.93% affirmative HR 7885. Foreign Assistance Act of 1963. Amend foreign aid law and authorize appropriations of $4,087,075,000 for foreign aid in fiscal 1964. Adair (R Ind.) motion to recommit the bill to the Foreign Affairs Committee with instructions to make the following cuts in authorizations: Alliance for Progress from $600 million to $450 million; military assistance from $1.225 billion to $1 billion; the development loan fund from $1.5 bil- lion to $900 million--a cut of only $160 million because Admin- istration asked appropriation of only $1.060 billion of Standing authorization; contingency fund from $200 million to $150 million. Motion agreed to 222-188. 1963 21 (Reflected) 47.45% affirmative HR 5517. Supplemental Appropriations for fiscal 1963. Lipscomb (R Calif.) motion to recommit the conference report with instructions that House conferees disagree to a Senate amend- ment providing $65,000 as a U. S. contribution to the Internation- al Peace Corps Secretariat. Agreed to 207-190. 1964:13 (Reflected) 47.65%.affirmative HR 9022. Authorize $312 million as the U. 5. contribution to an increase in the financial resources of the International Development Association. Talcott (R Calif.) motion to recommit the bill to the Banking and Currency Committee, Agreed to 208-189. 235 236 1964:61 (Reflected) 50.96% affirmative HR 11812. Rhodes (R Ariz.) motion to recommit the bill to the Appropriations Committee with instructions to reduce economic aid funds by $247.8 million as recommended by Rep. Passman (0 La.), chairman of the Foreign operations Sub-committee. Rejected 198- 208. 1964:51 (Reflected) 52.29% affirmative HR 11380. The Foreign Assistance Act of 1964. Adair (R 1nd.) motion to recommit the bill with instructions to reduce the fiscal 1965 authorization for development loans by $750,000,000 and the President's contingency fund by $50,000,000. Rejected 193-211. 1963:100 53.53% affirmative HR 7885. Foreign Assistance Act of 1963, authorizing appro- priations of $3,602,075,000 for foreign aid in fiscal 1964 and adding a number of legislative restrictions on the program. Adop- tion of the conference report. Adopted 195-164“ 1963:62 54.35% affirmative HR 7885. Authorize appropriations of $3,502,075,000 for foreign aid in fiscal 1964. Passage of the bill. Passed 224-186. 1964-52 56.59% affirmative HR 11380. Passage of the bill authorizing $2.041 billion in new foreign aid appropriations for fiscal 1965. Passed 230-175. 1964: 62 56.73%.affirmative HR 11812. Passage of the bill appropriating $3,316,572,400 for foreign assistance and $422,677,000 for related programs. Passed423l-l74. 1963: 112 63.73% affirmative HR 9499. Passage of the Foreign Aid Appropriation for fiscal 1964, appropriating $2,801,700,000 for foreign aid, $295,580,000 for other international programs, $2,838,275 for the House of Repre- sentatives and $13 million for claims against the U. S., as amended. Passed 250 to 135. A-2 237 1964: 40 (Reflected) 64.13% affirmative S 2214. ‘Authorize $312 million as the U. S. contribution to an increase in the financial resources of the International Develop- ment Association. Clawson (R Calif.) motion to recommit (kill) the bill. Rejected 132-247. 1963: 95 64.71% affirmative S 777. Authorize a two-year, $20 million appropriation for the Arms Control and Disarmament Agency. Passed 251-134. 1964: 86 71.02%.affirmative S 1627. Enable the U. S. to pay its share of fiscal 1964 operating expenses of the International Commission for Supervision and Control in Laos. Passed 268-89. 1964: 15 (Reflected) 76.77% affirmative S 2455. Authorize $115 million for Peace Corps operations in fiscal 1965. Gross (R Iowa) motion to recommit the bill to the Foreign Affairs Committee with instructions to reduce the authori- zation to $95,963,971--the total appropriated in fiscal 1964. Rejected 90-309. Foreign trade (3 items) CR = .985; CS I .958 1964: 96 48.69%.affirmative HR 8864. Enable the United States to implement enforcement of the International Coffee Agreement of 1962. Mills (0 Ark.) motion to adopt the conference report. Rejected 183-194. 1963: 90 54.57% affirmative HR 8864. Authorize the President to limit coffee imports and require certificates of origin from exporting countries.in accordance with the International Coffee Agreement of 1962, which was ratified by the Senate May 21, 1963. Passed 181-145. 1964:106 97.50% affirmative HR 12298. Passage of the bill extending PL 480 for three years, through Dec. 31, 1967, and authorizing the commitment of up to $4 billion for Title I over the three years, plus the use of unused previous authorizations, and $450 million annually for Title 11, plus unused previous authorizations. Passed 349-6. 8-2 238 Consumer protection (3 items) Cr = .993; CS = .949 1964: 14 (Reflected) 13.45% affirmative HR 8316. Amend the Communications Act of 1934 to prohibit the Federal Communications Commission from setting Standards govern- ing the length or frequency of radio and television commercials. Passed 317-43. 1964: 47 (Reflected) 42.18% affirmative HR 5130. McDade (R Pa.) motion to recommit (kill) the bill, which raised the maximum insurance on depositors' accounts in banks and savings and loan associations but did not contain Administration provisions tightening regulations affecting savings and loan asso- ciations. Recommittal agreed to 197-142. 1964: 46 64.81% affirmative HR 5130. Raise the federal insurance on a depositor's ac- count in banks or savings and loan associations to $20,000. Open rule (H Res 724) for debate on HR 5130 agreed to 218-115. Conservation (5 items) CR - .972; CS = .919 1963: 18 (Reflected) 14.56% affirmative HR 1762. Give statutory authority to the Interior Depart- ment's Bureau of Outdoor Recreation. Griffin (R Mich.) amendment to require the Secretary of Interior to report annually to Congress full details of land or other donations to the Government under the bill. -Agreed to 292-50. 1964: 74 53.41% affirmative HR 1096. Authorize the Secretary of the Interior to cooper- ate with the state of Wisconsin in designating and administering an Ice Age National Scientific Reserve. Aspinall (D Colo.) m0tion to pass the bill under suspension of the rules (two-thirds major- ity vote required). Rejected 164-154. 1964: 109 59.57% affirmative HR 1096. Authorize the Secretary of the Interior to cooper- ate with the State of Wisconsin in designating and administering an Ice Age National Reserve. Passed 180-118. 239 1964: 75 78.66% affirmative HR 3672. Authorize $47 million for construction of the Savory-Pot Hook, Bostwick Park and Fruitland Mesa irrigation pro- jects as part of the Colorado River Storage Project. Aspinall (D. Colo.) motion to pass the bill under suspension of the rules (two-thirds majority required). Passed 250-67. 1963: 92 90.97% affirmative HR 8135. Authorize up to $1.1 million to establish and administer public recreational facilities at the Sanford Reser- voir area of the Canadian River reclamation project in northwest Texas near Amarillo. (The $98 million Canadian River project, authorized in 1950, is scheduled for completion in 1967.) Passed under suspension of the rules 283-30 Agricultural policy (7 items) CR - .943; CS = .862 1963: 16 (Reflected) 50.84% affirmative HR 4997. rAuthorize a voluntary feed grains acreage diver- sion program for 1964-65 similar to those in effect for 1961-63. Harvey (R Ind.) motion to recommit the bill to the Agriculture Committee. Rejected l96-205. 1963: 17 51.18% affirmative HR 4997. Passage of the bill, authorizing a voluntary feed grains acreage diversion program for 1964-65. Passed 208-195. 1963: 99 54.18% affirmative HR 6196. Passage of the bill. Passed 216-182. 1963: 98 (Reflected) 55.61% affirmative HR 6196. Authorize the Government to pay a cotton subsidy in kind to domestic textile mills or cotton handlers in order to eliminate the competitive inequity between raw cotton prices on the world market and those on the domestic market; establish new price support levels for cotton; authorize a research program into cot- ton production costs; and establish an optional ”export market acreage allotment” for cotton farmers. Hoeven (R Iowa) motion to recommit the bill to the Committee on Agriculture (in effect, kill it). Rejected 179-224. 240 1963: 4 61.69% affirmative H J Res 284. Make supplemental appropriation of $508,172,000 for the Department of Agriculture in fiscal 1963. Passed 254-154. 1963: 34 77.20% affirmative HR 6754. Appropriate $5,979,457,000 to the Department of Agriculture for fiscal 1964. Passed 288-79. 1964: 44 82.35% affirmative HR 11202. Passage of the Agriculture Appropriation Bill for fiscal 1965. .Passed 311-64. Urban improvement (9 items) CR - .958; CS = .894 1963: 40 30.83% affirmative HR 6177. Increase from $32 million-$45 million in fiscal 1964 and thereafter the ceiling on authorizations for federal payments to the District of Columbia. Cohelan (D Calif.) motion to recommit the bill to the District of Columbia Committee with instructions that it be reported with a $53 million authorization. Rejected 99-237. 1963: 32 (Reflected) 43.75% affirmative HR 3496. Extend for two years, until June 1, 1965, the Reorganization Act of 1949. Brown (R Ohio) amendment to prohibit the President from creating a new executive department by reor- ganization plan. Agreed to 226-175. 1964: 3 (Reflected) 50.66% affirmative S 2265. Amend the 1956 Library Services Act to extend federal aid for library services to urban (as well as rural) areas, increase to $25 million in fiscal 1964 for grants to the states for construction of public libraries. Frelinghuysen (R N.J.) amendment to delete construction aid and aid to services in urban areas, and double the existing population and authorization limits on aid to rural areas. Rejected 179-183. 241 1964: 4 (Reflected) 51.98% affirmative S 2265. Martin (R Neb.) motion to recommit the bill to the Education and Labor Committee with instructions to deleted pro- visions for library construction aid. Rejected 174-188. 1964: 59 52.49%.affirmative HR 3881. Urban Mass Transportation Act of 1964. Passage of the bill authorizing federal matching grants to states and localities totaling $375 million over three years as the ”first installment” of a program to improve urban mass transportation service. Passed, 212-189. 1964: 58 (Reflected) 52.89% affirmative HR 3881. Oliver P. Bolton (R Ohio) motion to recommit the bill to the House Banking and Currency Committee with instructions to defer action until the House and Senate Banking and Currency Committees had studied the results of current federal mass trans- portation demonstration programs and the status of metropolitan transportation planning being carried out pursuant to a require- ment of the 1962 Highway Act. Motion rejected 190—214. 1964: 5 69.76% affirmative S 2265. Passage of the bill. Passed 254-107. 1963: 104 71.20% affirmative HR 6518. The Clean Air Act, to initiate and strengthen programs for the prevention and abatement of air pollution. Adoption of conference report. Adopted 273-109. 1963: 47 72.78% affirmative HR 6518. The Clean Air Act, to initiate and strengthen pro- grams for the prevention and abatement of air pollution. Passed 273-102. Negro rights (5 items) CR = .992; C5 = .970 1963: 96 (Reflected) 44.76% affirmative HR 9124. Revise the junior and senior Reserve Officers' Training Corps programs of the Army, Air Force and Navy. Hebert 242 (D La.) motion to pass the bill under suspension of the rules (two-thirds majority vote required). Rejected 177-154. 1964: 9 68.74%.affirmative HR 7152. Civil Rights Act of 1964. Passage of the bill to enforce the right to vote; prevent discrimination in access to public accommodations and public facilities; expedite public school desegregation; extend the life of the Civil Rights Commission for four years and broaden its powers; prevent discrimination in ad- ministration of federally assisted programs; prevent discrimina- tion based on race, color, religion, sex or national origin in employment and labor union membership; direct Census studies of registration and voting based on race, color and national‘origin; establish a Community Relations Service to mediate racial dis- putes; and permit the Attorney General to instigate or intervene in certain civil rights cases. Passed 290-130. 1964: 63 68.90% affirmative HR 7152. Civil Rights Act of 1964. Adoption of a resolu- tion (H Res 789) providing for House approval of the bill as amended by the Senate, thus clearing it for the President to sign into law. Resolution adopted 289-126. 1963: 72 76.00%.affirmative HR 3369. Private bill for the relief of-Mrs. Elizabeth G. Mason to which an amendment extending'the life.6f‘the.Civ11.Rights Commission for one year had been added by the Senate. Considered under suspension of the rules (two-thirds affirmative vote re- quired). Passed 265-80. 1963: 33 (Reflected) 84.10% affirmative Roosevelt (D Calif.)--Williams (0 Miss.) motion to adjourn the House. The purpose of the motion was to stop Reps. Lindsay (R N.Y.), MacGregor (R Minn.) and other Republicans from discussing civil rights legislation under previously approved special orders for House speeches. Rejected 53-276. Civil liberties (14 items) CR = .952; CS = .850 1963: 5 (Reflected) 4.98%.affirmative H Res 249. Authorize House Un-American Activities Committee to spend $360,000 in calendar year 1963, provided no investigations duplicate those being conducted by other House committees. Adopted 386-20. 243 I964: 66 (Reflected) 5.29%.affirmative HR 319. Permit a person who received mail which he considered ”morally offensive” to request the Postmaster General to prevent the sender from mailing more unsolicited material to him or his child- ren (considered under suspension-of—the-rules procedure, which requires a two-thirds majority for passage). Passed 326-19. 1963: 43 11.38%.affirmative HR 4897. Repeal the geographic limitation on the operation of the Sedition Act of 1917. Lindsay (R N.Y.) motion to recommit (kill) the bill. Rejected 40-339. 1963: 22 (Reflected) 11.46% affirmative HR 950. Amend the Internal Security Act of 1950 to provide a legislative basis for personnel security procedures of the National Security Agency, an intelligence arm of the Defense Department. Passed 340-40. 1963: 23 (Reflected) 15.81 affirmative HR 4274. Permit principals and teachers in D. C. public schools to use ”reasonable force” in disciplining students. (The bill would nullify a D. C. Board of Education rule prohibiting use of corporal punishment. The bill was supported by the D. C. Superintendent of Schools, opposed by a majority of the Board of .Education.) Passed 278-53. 1964: 84 (Reflected) 20.84%.affirmative HR 5990. Require organizations seeking funds for charitable purposes in the District of Columbia to satisfy the D. C. Commis- sioners that their solicitation would promote the public health or welfare and would not offend public morals. Passed 301-81. 1963: 59 36.29%.affirmative HR 7525. Amend crime laws affecting the District of Columbia to increase investigative powers of the police, nullify the Mallory and Durham rules of evidence, and provide minimum and mandatory penalties for certain crimes. Mathias (R Md.) motion to recommit (kill) the bill. Rejected 114-222. 244 1964: 100 (Reflected) 38.24% affirmative HR 11926. Open rule (H Res 845) for debate on HR 11926 con- cerning state legislative reapportionment (See RC 101). Reso- lution adopted 242-148. 1964: 101 (Reflected) 44.53% affirmative HR 11926. Deny the Supreme Court and lower federal courts jurisdiction over matters dealing with state legislative reap- portionment. ~Passed 218-175. 19631 111 (Reflected) 44.02% affirmative* HR 9499. Foreign Aid appropriation bill for fiscal 1964, appropriating $2,801,700,000 for foreign aid, $295,580,000 for other international programs, $2,838,275 for the House of Repre- sentatives and $13 million for claims against the U. S. (grand total: $3,113,100,370). Jensen (R Iowa) motion to recommit the bill and insert an amendment designed to bar the Export-Import Bank from guaranteeing credits to Communist countries or their nationals for the purchase of U. S. commodities. Agreed to 218-169. 1964: 105 (Reflected) 48.84% affirmative HR 12298. Agriculture Trade Development and Assistance Act of 1954 (PL 480). Findley (R 111.) motion to recommit the bill with instruction to add an amendment prohibiting sale of surplus U. S. farm goods under Title I of PL 480 to nations con- trolled or dominated by a Communist government or--as under exis- ting law--by ”the world Communist movement.“ (The amendment was aimed at Poland and Yugoslavia.) Accepted 183-175. 1964: 43 (Reflected) 50.00% affirmative HR 11202. Bow (R Ohio) motion to recommit the bill with instructions to amend it to prohibit the use of funds for export payments or export subsidies on agricultural products shipped to Communist nations. Rejected 186-187. 1963: 42 79.34% affirmative HR 3179. Appoint judges to the U.S. Court of Military *1963: 111 in the Civil Liberties scale was the only item in the Study which was not positioned by percent affirmative votes. See chapter 11. 245 Appeals for life tenure instead of lS-year terms. Passed 314-82. 1963: 7 94.68%.affirmative HR 4374. Authorize the President to proclaim Sir Winston S. Churchill an honorary citizen of the United States. Passed 378-21. Aid to education (5 items) CR - .949; C5 = .816 I964: 111 (Reflected) 67.7V% affirmative S 3060. Amend and extend for three years, until June 30, 1963, the National Defense Education.Act and amend and extend for one year, until June 30, 1966, federal aid to ”impacted” school areas. Goodell (R N.Y.) motion to recommit the conference re- port (H Rept 1916) with instructions to delete provisions ex- tending NDEA programs to include aid for the subjects of history, geography, civics and English, and to include a two-year exten- sion of impacted areas laws. Rejected 107-237. 1964: 113 69.88% affirmative HR 12633. Provide $1,117,196,068 in supplemental fiscal 1965 appropriations for new or expanded programs authorized in 1964, including activities under the Economic Opportunity Act, the Civil Rights Act and the National Defense Education Act. Mahon (D Texas) motion that the House amend the conference bill to permit District of Columbia participation in the program of aid to schools in federally impacted areas through use of pre- viously appropriated funds. Motion agreed to 211-78. 1963: 60 70.35%.affirmative HR 6143. Provide a five-year program of federal grants and loans for construction or improvement of higher education academic facilities and authorize $1,195,000,000 for three years. Passed 287-113. 1963: 85 71.47% affirmative HR 6143. Conference report on the 1963 Higher Education Facilities Act, providing a five-year program of federal grants and loans for construction or improvement of public and private higher education academic facilities and authorizing $1,195,000,000 over the first three years. Agreed to 258-92. 246 1963: 53 88.79% affirmative S 1652. Amend the National Cultural Center Act to extend the fund-raising period for three years, through Sept. 2, 1966, and to increase the number of trustees from 15 to 30. Passed under suspension of the rules (two-thirds vote needed for approval) 293-33. Orientation to professionalism (6 items) CR I .944; CS = .825 1963: 14 (Reflected) 57.76% affirmative HR 12. Health Professions Educational Assistance Act of 1963. Devine (R Ohio) motion to recommit the bill to the Inter- state and Foreign Commerce Committee with instructions to report it back after deleting provisions for loans for medical, dental and osteopathic students. Rejected 171-239. 1964: 53 59.95% affirmative HR 11049. Raise the salaries of approximately 1.7 million federal career employees and of federal executives, judges and Members of Congress. Passed 243-157. 1964: 18 62.87% affirmative HR 8986. Raise the salaries of approximately 1.7 million federal career employees and of federal executives, judges and Members of Congress. Adoption of an open rule (H Res 650) for debate on the bill. Adopted 251-147. 1964: 72 65.24% affirmative H Res 803. Resolution disagreeing to Senate amendments and calling for a conference on.a bill (HR 11049) to raise the salaries of approximately 1.7 million federal career employees and of fed- eral executives, judges and Members of Congress. Resolution adopted 245-131. 1963: 15 69.66% affirmative HR 12. Passage of the bill, authorizing a three-year pro- gram of matching grants for construction and rehabilitation of teaching facilities for medical, dental and related professional schools and providing a six-year loan program for students of medicine, dentistry and osteopathy. Passed 288-122. 247 1963: 80 88.15% affirmative HR 5945. Establish a United States-Puerto Rico Commission on the Status of Puerto Rico. Passed 320-44. Labor versus business (8 items) CR I .961; CS = .888 1963: 36 49.52% affirmative HR 4996. Area Redevelopment Act Amendments of 1963. Re- jected4204-209. 1964: 55 (Reflected) 52.83% affirmative HR 11376. Extend for one year, through June 30, 1965, $1.9 billion in Korean War excise tax rates. Byrnes (R Wis.) motion to recommit the bill to the Ways and Means Committee with instructions to eliminate, over two years, certain World War 11 retailers' excise taxes. Rejected 185-207. 1963: 68 (Reflected) 53.04% affirmative HR 8363. Byrnes (R Wis.) motion to recommit the bill to the Ways and Means Committee with instruction to amend it to prevent the tax reductions from taking effect unless President Kennedy specified that administrative budget spending for fiscal 1964 was not expected to exceed $97 billion and for fiscal 1965 was not expected to exceed $98 billion. Rejected 199-226. 1963: 12 55.02% affirmative HR 5517. Supplemental Appropriations for fiscal 1963. Boland (0 Mass.) amendment to add $450 million for the 1962 accelerated public works program. Accepted 228-184. 1964: 57 57.92%.affirmative HR 3881. Urban Mass Transportation Act of 1964, authorizing $375 million over three years to initiate a program of federal matching grants to states and localities to build and improve mass transportation systems. Rains (0 Ala.) amendment broadening the job protection afforded under the federal aid by (l) guaranteeing the preservation of all employee rights (such as vacation time, pension rights, etc.) whether or not acquired through collective bargaining, (2) guaranteeing that employees of a private mass tran- sit company acquired under the bill would become employees of the new company, maintaining all their old rights, and (3) extending 248 the guarantees against worsening of position afforded railroad workers affected by mergers or acquisitions under the Interstate Commerce Act (Section 5 (2) (f)) to all transit workers. Accepted 234-170. 1963: 69 63.70%.affirmative HR 8363. Passage of the bill, lowering personal and cor- porate income taxes and making other changes in the Internal Revenue Code of 1954. Passed 271-155. 1964: 12 79.51% affirmative HR 8363. Conference report on the Revenue Act of 1964, reducing personal and corporate income tax liabilities by $11.5 billion over a two-year period, lowering personal income tax rates from a range of 20 to 91 percent to a range of 14 to 70 percent, lowering personal income tax liabilities by an average 19.4 per- cent, reducing the corporate income tax rate from 52 to 48 percent in a manner giving special benefit to small business and making other structural changes in the Internal Revenue Code of 1954. Accepted 326-83. 1963: 67 84.72%1affirmative H Res 527. A closed rule preventing floor amendments and allowing one motion to recommit a bill (HR 8363), the Revenue Act of 1963. Adopted 324-67 Spending (14 items) CR I .941 C5 = .842 I963: 49 (Reflected) 2.83% affirmative HR 3872. Extend for five years, through June 30, 1968, the life of the Export-Import Bank of Washington and increase its lending authority. Patman (D Texas) motion insisting that House conferees disagree to a Senate amendment continuing ”backdoor financing” for the agency. Agreed to 379-11. 1964: 34 (Reflected) 42.52 affirmative HR 10723. Legislative appropriations bill for fiscal 1965, appropriating $173,626,640 to the House of Representatives and related offices. ~Lipscomb (R Calif.) motion to recommit the bill to the Appropriations Committee with instructions to report it back with an amendment, limiting expenditure by Congressional committees or the Capitoerrchitect to activities that are a matter of public record. Agreed to 188-131. 249 I964: 107 47.09%.affirmative S 2220. Authorize forgiveness of up to 50 percent of student loans plus accrued interest made under the 1963 Health Professions Educational Assistance Act (PL 88-129). for physicians, dentists, and osteopaths subsequently practicing in an area in which there was a certified shortage of health personnel. Rejected 140-161. 1963: 77 50.27% affirmative HR 6237. Authorize appropriation of $500,000 annually for five years to help public and private groups compile and publish documentary source materials significant to the history of the United States. Passed 158-154. 1963: 9 (Reflected) 62.69%.affirmative HR 2440. Curtis (R Mo.) motion to recommit the bill with instructions to reduce authorizations for all items except.Air Force research and development and Navy ship and torpedo procure- ment by 5 percent--a cut of $636,385,250. Rejected 149-258. 1963: 24 63.59% affirmative HR 5517. Supplemental Appropriation bill for fiscal 1963. Thomas (D Texas) motion to adopt the conference report. Agreed to‘24l-l30. I963: 35 67.76% affirmative HR 6868. Legislative Appropriation bill for fiscal 1964, appropriating $140,038,919. Passed 271-122. 1963: 45 86.58%.affirmative HR 5279. Adoption of conference report appropriating $958,456,500 for the Interior Department and related agencies in fiscal 1964. Agreed to 332-50. 1963: 75 89.47% affirmative HR 8747. Independent Offices Appropriation bill for fiscal 1964, providing $13,102,718,7OO for 23 executive agencies and independent offices. Passed 302-32. 250 1963: 102 94.15%1affirmative HR 8747. Independent Offices Appropriations for fiscal 1964. Adoption of conference report appropriating $13,224,518,050 for 23 executive agencies and independent offices. Agreed 356-22. 1964: 35 94.29% affirmative HR 10908. Provide $6,908,063,000 in fiscal 1965 appropria- tions for the Department of Labor and Health, Education and Wel- fare and related agencies. Passed 347-21. 1963: ll 95.71%.affirmative HR 5366. Appropriate $5,997,026,000 to the Treasury and Post Office Departments, the Executive Office of the President and three independent agencies in fiscal 1964. Passed 385-17. 1963: 93 98.52%.affirmative HR 9139. Military Constructing Appropriation bill for fis- cal 1964 appropriating $1,562,964,000 for construction at military bases in the U. S. and abroad and for payments for the family housing program. Passed 332-5. 1963: 73 99.11% affirmative HR 7179. Conference report appropriating $47,220,0I0,000 to the Defense Department during fiscal 1964. Adopted 336-3. Space spending (6 items) CR = .983; CS = .946 1963: 63 52.50% affirmative HR 7500. Fiscal 1964 National Aeronautics and Space Adminis- tration authorization. Pelly (R Wash.) motion to recommit the conference report (H Rept 706) with instructions to reduce the authorization to the $5,203,719,4OO amount approved by the House. Rejected 177-200. 1963: 74 56.46% affirmative HR 8747. Independent Offices Appropriation bill for fiscal 1964. Wyman (R N. H.) motion to recommit the bill to the Appro- priations Committee with instructions to reduce by $200 million the $3.9 billion appropriated to the National Aeronautics and Space Administration for research and development. Rejected 145-192. 251 1963: 64 65.48% affirmative HR 7500. Adoption of the conference report authorizing $5,350,820,4OO for the National Aeronautics and Space Adminis- tration in fiscal 1964. Adopted 248-125. 1964: 87 (Reflected) 70.37% affirmative HR 11296. Appropriate $13,454,859,000 in fiscal 1965 for 16 independent agencies, three small agencies in the Executive Office of the President, the President's disaster relief fund and civil defense activities of the Health, Education and Welfare and Defense Departments. Wyman (R N.H.) motion to recommit the bill with instructions to reduce funds for the National Aeronautics and Space Administration by $200 million. Rejected 114-271. 1964: 26 79.72%.affirmative HR 10456. Passage of the bill, authorizing $5,193,810,500 for the National Aeronautics and Space Administration in fiscal 1965 in the following categories: research and development, $4,327,950,000; construction of facilities, $248,335,000; adminis- trative operations, $617,525,500. Passed 283-73. 1963: 52 85.42% affirmative HR 7500. Authorize $5,203,719,4OO for the National Aero- nautics and Space Administration in fiscal 1964. Passed 335-57. Orientation to debt (9 items) CR = .979; CS = .952 1963: 89 50.97% affirmative HR 8969. Passage of the bill to increase the temporary national debt limit to $315 billion through June 29, 1964. Passed 187-179. 1963: 28 51.07% affirmative HR 6009. Passage of the bill. Passed 213-204. 1964: 56 52.67% affirmative HR 11375. Increase the temporary national debt limit to $324 billion through June 30, 1965. Passed 203-182. 252 1963: 88 (Reflected) 53.01% affirmative HR 8969. Extend the existing temporary $309, billion national debt limit from Nov. 30 through June 30, 1964 and further increase the temporary limit by $6 billion from Nov. 30 through June 29, 1964. Byrnes (R Wis.) motion to recommit the bill to the Ways and Means Committee in order to establish a smaller increase in the ceiling for the remainder of fiscal 1964. Rejected 172-197- 1963: 27 (Reflected) 53.19% affirmative HR 6009. Increase the temporary national debt limit to $307 billion from time of enactment of bill through June 30 and to $309 billion from July through Aug. 31. Byrnes (R Wis.) mo- tion to recommit the bill to the Ways and Means Committee with instructions to amend it by extending the existing $305 billion ceiling indefinitely. Rejected 195-222. 1963: 58 55.49% affirmative HR 7824. Passage of the bill extending the existing tem- porary limit for three months, from Aug. 31 to Nov.30, 1963. Passed 221-175. 1963: 87 57.52% affirmative H Res 564. A closed rule, preventing floor amendments and allowing one motion to recommit, for consideration of HR 8969, raising and extending the temporary debt limit. Adopted 212-149. 1963: 57 (Reflected) 57.82%.affirmative HR 7824. Extend the existing temporary $309 billion national debt limit. Byrnes (R Wis.) motion to recommit the bill to the Ways and Means Committee with instructions to amend it by reducing the temporary limit to $307 billion from $309 billion and to extend it for two months, from Aug. 31 to Oct. 31, 1963. Rejected 164-229. 1963: 56 _ 79.09%.affirmative HR 7824. Adoption of a closed rule (HR Res 477) preventing floor amendments to the bill extending the existing debt limit. Adopted 303-72. .%—. 253 Social welfare (15 items) CR = .956; C5 = .873 1963: 108 (Reflected) 51.16% affirmative HR 4955. Authorize new funds for an expanded vocational education program, and extend the National Defense Education Act (NDEA) and aid to federally impacted school areas. Authoriza- tions for vocational education and NDEA programs totaled $921 million over a four-year period. Frelinghuysen (R N.J.) motion to recommit the conference report with instructions to delete authorizations of $150 million over four years for work-study programs and residential vocational education schools. Recom- mittal motion rejected 180-193. 1964: 79 (Reflected) 52.73%.affirmative HR 11377. The Economic Opportunity Act of 1964, authorizing $947.5 million in fiscal 1965 for a variety of programs to combat poverty. Smith (D Va.) motion to strike the enacting clause (kill the bill). Defeated 197-225. 1964: 29 (Reflected) 53.00% affirmative HR 10222. Hoeven (R Iowa) motion to recommit the bill and add a provision requiring the states to pay half the costs of food stamp programs set up under the bill. Rejected 195-223. 1964: 80 53.94% affirmative HR 11377. Landrum (D Ga.) substitute amendment for HR 11377, embodying the Senate-passed bill plus two major changes: it pro- vided a veto power to state Governors over public and private projects under the community action program, and deleted authority for the Director of the poverty program to cancel repayment of cer- tain loans. Accepted 228-190. 1964: 30 54.33% affirmative HR 10222. Food stamp bill, authorizing the Secretary of Agriculture to set up and finance state and local food stamp programs, and authorizing $400 million over fiscal years 1964-67 to cover the costs. Passage of the bill. Passed 229-189. 1964: 82 54.69% affirmative S 2642. Passage of the anti-poverty bill which incorporated the text of HR 11377 (above). The bill, the Economic Opportunity Act of 1964, authorized for three years, with spending of $947.5 254 million in fiscal 1965, a variety of programs to combat poverty. Passed 226-185. 1963: 50 58.42%.affirmative HR 5207. Foreign Service Buildings authorization for fiscal 1964 and 1965. Adoption of a special rule waiving points of order against language in the conference report that added to the bill a Philippine war claims amendment. Adopted 234-166. 19641 99 80.77% affirmative S 3049. Housing Act of 1964, authorizing $1,130,750,000 to fund new and existing housing and urban renewal programs through Sept. 30, 1965. -Adoption of the conference report. Adopted 310-700 1963: 109 81.02%.affirmative HR 4955. Adoption of the conference report. Agreed to 301-65. 1964: 89 81.36%.affirmative HR 12175. Passage of the Housing Act of 1964, authorizing $992 million to fund new and existing programs through June 30, 1965. Passed 308-68. 1964: 70 92.08% affirmative HR 11865. Social Security Amendments of 1964. Adoption of a closed rule limiting debate to five hours and prohibiting floor amendments. Adopted 350-29. 1963: 65 94.33% affirmative S 1576. The Mental Retardation Facilities and Community Mental Health Centers Construction Act of 1963, authorizing a three-year program of $238 million in grants to states and public and private institutions to combat mental illness and retardation. Passed 335-18. 1963: 78 95.50% affirmative S 1576. Mental Retardation Facilities and Community Mental Health Centers Construction Act of 1963, authorizing $329 million 255 over a four-year period. Adoption of the conference report. Agreed to 299-13. 1964: 71 97.97% affirmative HR 11865. Passage of the bill, providing a 5 percent across-the-board increase in Social Security old-age, survivors and disability benefits; raising the wage base for the Social Security tax from $4,800 to $5,400 a year; increasing payroll and self-employed tax rates; permitting widows to receive benefits at age 60 instead of 62; and bringing self-employed doctors and interns under compulsory coverage. Passed 388-8. 1963: 115 98.79% affirmative H J Res 875. Appropriate $41,886,000 in fiscal 1964 for the Department of Health, Education and Welfare to implement two mental retardation bills (PL 88-156 and PL 88-164) passed in 1963. Passed 325-4. Party loyalty (14 items) CR = .981; CS = .953 1964: 91 44.62%iaffirmative H Res 663. Appropriate $10,000 to the House Education and Labor Committee to cover expenses incurred by its Ad Hoc Poverty War Pragram Subcommittee in its work on President Johnson's anti- poverty bill. Rejected 115-156. 1964: 88 (Reflected) 51.13% affirmative HR 12175. Housing Act of 1964. Kilburn (R N.Y.) motion to recommit the bill to the Banking and Currency Committee with in- structions to report it with an amendment extending indefinitely the insurance authority of the Federal Housing Administration beyond the expiration date of Oct. 1, 1965. Rejected 184-194. 1963: 119 53.73%.affirmative HR 9499. Passman (D La.) motion to adopt an amendment permitting the President to authorize Export-Import Bank guaran- tees on credit to Communist countries for purchase of U. S. com- modities if he considered it in the national interest and notified Congress within 30 days after each such determination. Agreed to 189-158. 256 I963: 103 (Reflected) 54.55%.affirmative HR 8747. Ostertag (R N.Y.) motion to concur in a Senate amendment requiring the Administrator of the Veterans Adminis- tration to use a previously appropriated and earmarked $1,722,000 for construction of an addition to the veterans hospital at Bay pines, Florida. Rejected 171-204. 1964: 8 55.70% affirmative HR 7152. Civil Rights Act of 1964. Albert (D Okla.) mo- tion that the House adjourn (at 10 p.m.) until Monday, Feb. 10, rather than move immediately to consideration of final titles of the bill and a vote on passage. Agreed to 220-175. 1963: 106 56.43% affirmative Motion by Majority Leader Albert (0 Okla.) to adjourn the House rather than proceed to consideration of bills under Calen- dar Wednesday procedure. (Republicans had announced they would attempt to bring up HR 7152, the civil rights bill, under the Calendar Wednesday rule.) Motion agreed to 214-166. 1964: 28 (Reflected) 58.25%.affirmative HR 10222. Food stamp bill. Oliver P. Bolton (R Ohio) mo- tion to adjourn for the day without completing action on the bill. Rejected 173-239. 1964: 97 58.84% affirmative HR 8000. Interest Equalization Tax Act, imposing a temporary, retroactive tax on the purchase by Americans of certain foreign securities in order to restrict foreign borrowing in U. S. capital markets. Adoption of the conference report. Adopted 221-147. 1963: 1 59.24% affirmative Election of Speaker of the House of Representatives for the 88th Congress. The nominees were Reps. John.W. McCormack (0 Mass.), Speaker since January 1962, and Charles A Halleck (R Ind.), Minority Leader for the first time in the 86th Congress. McCormack was elected, 256-175. 257 1964: 16 61.69% affirmative HR 8000. Interest Equalization Tax, imposing a temporary tax on the purchase by_Americans of certain foreign securities in order to restrict foreign borrowing in U. S. capital markets. Passed 238-142. 1964: 32 62.92% affirmative Albert (0 Okla.) motion to dispense with further proceedings after a quorum call in the House. (”Further proceedings” are the locating by the Sergeant at Arms of Members who did not answer to the quorum call. This is routinely dispensed with. Some Repub- lican Members, however, demanded the roll call in order to delay the Democratic leadership's schedule.) Agreed to 223-132. 1964: 33 65.64% affirmative Motion to dispense with further proceedings after a second Republican-demanded quorum call. Agreed to 234-122. 1963: 118 65.99% affirmative HR 9499. Motion to consider a rule (H Res 600) waiving points of order on the second conference report on the foreign aid bill (two-thirds majority necessary to consider the rule be- fore it had lain on the Speaker's desk for 24 hours). Rejected 202-105. 1964: 81 (Reflected) 71.02% affirmative HR 11377. Frelinghuysen (R N.J.) motion to recommit the Economic Opportunity Act, as amended, with instructions to the Education and Labor Committee to report it amended by a Repub- lican substitute bill. Motion rejected 117-295. Constituency data intercorrelations. APPENDIX C The entire table of simple correlations (Pearson product-moment) is given first, followed by a set of summary information reporting the correlations (rounded to two places) t .65 or higher. The data are nominally categorized for convenience. Finally, there is a list of reciprocal pairs; two variables each of whose highest correlate is the other. Variable Name % Population change 1 % Urban 2 % Negro 3 -219 030 % Foreign stock 4 081 639 -330 Median age 5 -193 421 -213 577 % With private elementary _ education 6 -O48 611 -207 713 519 %.With low education 7 -302 -316 617 -353 -412 % With high school education 8 546 415 -493 347 172 % With college education 9 415 431 -224 306 187 Median education 10 512 484 -477 396 218 Median income 11 468 710 -414 637 355 % Unemployed 12 -260 -109 222 -008 -019 % Owner occupied dwelling units 13 359 -374 -427 -343 -284 .% Sound units with all plumbing I4 500 770 -368 646 415 Median rooms per dwelling unit 15 157 -125 -304 -O38 -O38 Median persons per dwelling unit 16 335 I241 015 -224 -715 Median home value 17 397 720 -l90 694 380 Median rent 18 522 650 -323 610 310 % White collar workers 19 401 699 -304 556 415 % Blue collar workers 20 -202 022 294 -119 -843 % Farmer 21 -232 -746 035 -467 -360 I 2 3 4 5 % Popu- % Ur- % Ne- % For- Median lation ban gro eign age change Stock 258 % With private elementary 259 education 6 % With low education 7 '352 % With high school education 8 I74 -733 % With college education 9 I94 -394 804 Median education 10 226 -783 956 724 Median income 11 565 -687 754 656 785 % Unemployed 12 -138 247 -376 -406 -330 % Owner occupied dwelling units 13 -239 -311 219 -005 180 % Sound units with all plumbing 14 565 -685 709 547 758 Median rooms per dwelling unit 15 121 -39O I72 O31 175 Median persons per dwelling unit 16 -l79 244 -102 -O91 -l38 Median home value 17 569 -467 636 663 645 Median rent 18 520 -657 746 619 753 % White collar workers 19 459 -514 778 839 757 % Blue collar workers 20 -012 243 -520 -546 -409 % Farmer 21 -466 308 -320 -362 -402 6 7 8 9 10 % With % With % With % With Median private low high col- edu- elemen- educa- school lege cation tary tion edu- edu- educa- cation cation tion Median income 11 % Unemployed 12 -276 % Owner occupied dwelling units I3 047 -217 % Sound units with all plumbing 14 909 -290 053 Median rooms per dwelling unit 15 234 -260 674 Median persons per dwelling unit 16 -020 -147 461 -090 439 Median home value 17 854 -303 -243 781 -001 Median rent 18 904 -260 005 843 179 % White collar workers 19 811 -340 -106 772 -011 % Blue collar workers 20 -l68 426 -104 -148 034 %.Farmer 21 679 -042 207 -662 -020 11 12 13 I4 15 Median %.unem- % Own- % Median Income ployed er sound rooms occu units per pied with dwel- dwel- ,all ling ling plumb- unit units ing Median persons per dwelling unit Median home value Median rent % White collar workers % Blue collar workers % Farmer 16 17 18 19 20 21 260 -083 -035 -216 124 112 16 Median persons per dwel- ling unit 830 776 -251 '573 17 Median home value 761 -284 -480 -523 -588 -41 6 I8 19 20 Median % % rent White Blue collar col- worker lar work- ers 261 Demographic variables: 1 %.Population change, 2 %.Urban, 3 % Negro 4 % Foreign stock, 5 Median age % Population change % Urban 1 2 [None; the highest is 11 Median income = .71 8 %.with high school 14 % Sound units with all education = .55] plumbing = .77 17 Median home value = .72 18 Median rent = .65 19‘% White collar workers = .70 20 % Farmer = .75 % Negro 3 [None; the highest is 7 % With low education = .61 8 % With high school education =-.49 11 Median income =-.42 13 %.Owner occupied dwel- ling units =-.43] % Foreign stock Median age '1 5 6 %.With private elementary 16 Median persons per dwel- education = .71 ling unit =-.72 14 %.Sound units with all plumbing = .65 17 Median home value I .69 262 Education variables: 7‘% With low education, 8 %.With high school educa- tion, 9‘% With college education, 10 Median educa- tion % With low education % With high school education 7 8 8 % With high school educa- 7 %.With low education I-.73 tion I-.78 9 %.With college educa- 10 Median education I-.78 tion I .80 11 Median income I-.68 10 Median education I .95 14 % Sound units with all 11 Median income I .75 plumbing I-.68 14 %.Sound units with 18 Median rent I-.65 all plumbing I .71 18 Median rent = -73 % With college education Median education 9 10 8 % With high school educa- 7 % With low education I-.78 tion I .80 8 % With high school 10 Median education ,I .72 education I .95 11 Median income I .66 9‘% With college edu- cation = .72 17 Median home value I .66 11 Median income = .79 19‘% White collar workers I .84 I4 % Sound units with all plumbing I .75 18 Median rent = .75 19 %.White collar workers .76 263 Economic variables: 11 Median income, 14 %.Sound units with all plumbing, 17 Median home value, 18 Median rent Median income % Sound units with all plumbing ll 14 2 % Urban I .71 2 % Urban I .77 7 % With low education I-.69 4 %.Foreign stock I .65 8 % With high school 7 %.With low education I-.68 education = .75 8 % With high school 9 %.With college education I .65 education I .71 10 Median education I .78 -.10 Median education I .76 14 %iSound units with all 11 Median income I .91 plumbing = .91 17 Median home value I .78 17 Median home value I .85 18 Median rent I .84 18 Median rent I .90 ' 19'% White collar workers I .77 19 % White collar workers I .81 21 % Farmer I-.66 21 % Farmer I-.67 Median home value Median rent 17 18 2 % Urban I .72 2 % Urban I .65 4 % Foreign stock I .69 7‘% With low education I .66 9 %.With college education I .66 8‘% With high school education I .73 11 Median income I .85 10 Median education I .75 14 %.Sound units with all plumbing I .78 11 Median income I .90 18 Median rent I .83 14*% Sound units with all plumbing I .84 19 %.White collar workers I .78 17 Median home value = .83 19‘% White collar workers .76 264 Employment variables: 12 %.Unemployed, 19 % White collar workers, 20‘% Blue collar workers, 21 % Farmer % Unemployed % White collar workers 12 19 [None; the highest is 20 % Blue . 2 % Urban I .70 collar workers I .43] 8 % With high school education I .78 9 % With college education I .84 10 Median education I .75 11 Median income I .81 14 % Sound units with all plumbing I .77 17 Median home value I .78 18 Median rent I .76 % Blue collar workers %.Farmer 20 21 [None; the highest is 9 % With 2 % Urban I-.75 college education I-.55] 11 Median income I-.68 l4 % Sound units with all plumbing =-.66 Sundry remaining variables: 6 %1With private elementary education, 13 %.Owner occupied dwelling units, 15 Median rooms per dwelling unit 16 Median persons per dwelling unit % With private elementary % Owner occupied dwelling units education 13 6 15 Median rooms per dwel 4 %.Foreign stock I .71 ling unit = .67 Median rooms per dwelling unit 15 l3 % Owner occupied dwelling units Highest reciprocal correlations: 4 % Foreign stock 5 Median age - 8 %.With high school education 9 %.With college education 11 Median income 13 %.Owner occupied dwelling units 265 Median persons per dwelling unit 16 5 Median age I-.72 I .67 six pairs .71 6 % With private elementary education .72 16 Median persons per dwelling unit .95 10 Median education .84 19 %.White collar workers .91 14 % Sound units with all plumbing .67 15 Median rooms per dwelling unit 266 3. gm .mi. 0; Sm a .31.. 1.04 ..~ 08 NM 3.5 out: 8.95.. 3. .3 .:o :8 :8 .3 ...N 62 dd .3 .mo .3 8:82.... 3 a... we. a a «H i. B. um. 02 9. SN dd 00 22.32339... 3 8:38.... a .... 08 m4 mom NS mac .3 mm 08 .19 .m-.. 8.68% puma... $1.. a a .:N 03 ms 08 go SN 8. mm... 3... Soc 8 8.6.38.5 N... m2 8. Wm .flm cm: :2 mm: am 0% MS .3 3cm... otmoz as. .No omo 0.0 mN. m0. ~m. .mN 0o. omN oNo ..-m. co.uuobota totancou 51M 1.1.. NM NM :3 H0 a 31.. SN .8 H G. 6:2. 3.3.3.3.. Nd a a .8 me So one 8.. o: S. 3. ..-: $652.. as LSee. $0 .0le a mloiw SN 30 mmo So mm. 2. ml... .0 3.96. been. 3o 8. N00 8. mom mm. mm. 0: 08 m: S. .3 uterus .268 8. N0 .8 N3 mm. am. S. 9% .8 men mum ..-... 6:8 8.88 m... 03 .90. mm m... ewe .3 m... 86 a... a .N-... 839628 to «we: ma H 0: ..m. N... m: :8 NS mm .3 68863.... coat: «NM “01. a :8 son as .3 m3 8. an. 1.0... .N-: 8.22.... .3. «Mo. ....a mum co. mm. mm. EN 3.. m3 8. .... mutton... :36 m. m. m M. m. ..1.. m .1... m. u. m. m .m m. .w .m U U 3 U. 1. U. .4 U. J U. U a J E a n I. l. 1. IA 1. O U 1. I. a O 3 O U. 0 I d E 5 E U D. U 0 U I. U 0 8 J 5 U 1. 3 n l. ..D M D. I. 9 I. 0 3 ..l u. n A S 0 w 9 3 3 E 1. U a 1. 5 S 9 1. O I. 3 3 .4 3 . 3 O U. I. ua. U 0 O 3 O U I. l ... H” mm ,6 no I. o; C. h. C. 7. .l muu.cum.o ..m use mconeoe ..miimco.m ices.o 0c.uo> ..mu ..0L 0. new mo.nm.cm> >uc00u.umcoo ucoocoaooc. .N 00 mco.um.oLLoo :mhii.0 x.0zmmm< 267 N3 :8 a a a film wwlN. HNI. wolN m: .N-... 2:... .380“. a .mo 0... N: mm. .mINI. Hum S. .. o S. .... 8:83.... o. u... a on. o... NN.. mm. ...... .... .N. MNIN ...: .... Em........5.$30... 3 8:35.... SN 2. .N.. mm 8.. a PIN 8.. mm S. a-.. 3.22... 82.. "NI. .m. a «mu. :4... m... «.lN .Imm NWN. .3. ...-.. 32. o... 8:35.... a 9... RN ...m 93 m..|. E. ..R .mo 8.. a. 3...... 0...»... “fl 0... R. mu... om. «mm fl... mm. a m... ..-... 8.888.. 32:28 MIG NN. dill. ..N m a ..IN... .N...|N a mm. 8. 5:2. 3.3.3.3.. N... NN. ..No a E ...... a NI... 3 RN ..-: 32.3.. m> .33 .Nlml... .3. a m3 .8 3m HM NN.. SIN N..N .... 3.96. 5...... ....N NM. M... m... ..3 Nola a 8.. m AK. .3 83.2. .....oom 9N M... ..m. 8. R. E ..NMH ...: .:...m 3. ...<. 5:8 ......8... ...fl 8.. 9.. N... ...o ..o. .N..N.m 93 ..NN ooN .N-... 839.38. a ..m. E +3.. 3.. «mm o . 3.. and ..NN 3. 2053...... :3... 3|. 3. NS 93. mum mam NM «mm mlmw. om. .N.... 2.2.25 «mm m... ...: N... N... «am and NN. mm o: . 83.2... :3. % m % m % w w my w my w .....% 9% % N. wm mm m. w mm. nw. ...... mm m J 3 n 3 I. D. E I 9 IE .0 n IU 3 w J 3 J .4 U U I. U I. U I. U I. 3 w 3 S 5,3 U U n p U J .0 J m 9 M“ w. .b.w co m w.n s.n. m. u m. w ... m .N. m m m. m. m n a m .1. A u. w Us 5 a um p J D. S d 5 I. I a M 9 m w .. u. p J U: n M a. m U. m, m m 9. u ucoE_u mc_uo> __mu ..0. uuac.ucouunmuu..um_u __m vcm m.mnEoE __muumco_m 0. new mm_nm_.m> >ucm:u_umcou acovcoaouc.._~ we mco_um.o..ou amhun.o x_czumm< 268 moo. u .o.oa>.o .. mom. a .o,ou>.o .. O D 0": mn.: . _ m: on N xcmx MI. Nam. an. Non Nu. mad .o. co..mu=oo o. o.< . mow. o. on: . oNo ... m.;o.. o.moz m md. w mum N mmm. A_umv co_uuouo.m .osachU : on. m me 0 mm. as. m.mm_oz _m_u0m m mm. N mmm m WNW Ac. acmeo>o.aE_ cmn.= m :N_ m omm m mm. Au. mo_u.on__ __>_u. Am a Na. mco_mcus_ao >0..on. .~.uom on o. mN w .:a. MA N_m ml. MNN _MI. mlfl .Amu_v mc_ucmam vumam m mmN N N_N N mNH. A:u.v anon cu co_umuco_.o N mmN m ONN m _NHM A_u_v mmoc.man .m> .onm. o moN . .m. o. «MM AN-.. oo.oooom oxcmx m. xcmm m. axon. m. o.moz X cmn.: N. omcmsu. A: n .:. mco_mcoE.a m N co.um_:aom.$ >u__om u.Eocoum . o.nm..m> ucuucoaovc_ name .0» mco_mcme_n _m_00m vcm u_EOcouo co mco_um_o..ou mm on» mc..maeoonnmumou : >0:u_czuccmzuu.m x_ozmmm< 269 moo. u .o c.:.... .. moo. u .o cu>.o .. moo. u .:.ou>.o .. on: on: on: m: m: m: w xcmm NI mom .NI mmN .ml Nmm Ac. co_umo:uo Cu v.< o. mwm o. cm: 0. m:o Au. mu:m_. OLmoz m —NN m 3N. m mam A_umv co.uuuuo.m .053mcou m N_N m Nm_ m mdN Aw. «Law—03 _m_00m N mNN N OON m w.m An. acoeu>o.ae. cmn.: m mNN m NoN N NNm Am. m03.2... ..>.o Am u ch mco.mcue.o >u_.0m _m.00m o. o. o. w xcmz an. o.o .nu MNo Nu. moo Am-.. oo.ocmom oumom m 0:0 _ Nmo . mmo A¢u_v unov 0» co_umucu..o : oNo : N__ : ... A_u_v mmoc.msn .m> .onm. N oNo N moo N NNo .N-.. oo.ocoom .:m. m. xcmx m. .omx m. co.umo:vu >.Mucmeo_0 0mm cm_uoz x00um cm.o.ou x A: u _cv mco_mcoe.o. oum>_.a Lu.) x m : >u__om u.EOcoum o vu:c_ucouuuo_nm_.m> ucoucoaovc_ 50mm .0» m:o_mcue_n _m_u0m vcm u_eocouu co mco_um.u..ou mm on“ mc..maeoonumumou : >ocu_;3uccmznn.m x_ozumm< 270 m.o. u ..o cn>.o .. moo. u ..o cn>.o .. moo. u ..o c.:.... .. N u .3 0 fl .3 o n .3 MN _N _N N xcmx MI m_o MI. mmo MI. :00 Nov co.umu:vo cu u_< _ NM. _ NmN . mm“. Amy muzm_. o.moz N :mo m Mmm : NN. A_Imv co_u00uo.a .oeamcou m mum o wad. m oNN .o. n.n..nz .n.uom N mum m mwd. m :oN Nov ucoeo>o.ae_ cmn.= x mmm. : NNN N No. ... mo...no.. ..>.o Am u ch mco_mcue_a. >u..o. .n.uom N. .. .. n. ...... NI. Hum, NI. mHM NI. m_: AMI—v mc.ucoam oumam m a o. mum m man 3-: 3m... o. 5:33.... o. wmw. m ANN N mwN A.I_v mmoc.m:n .m> .onmq m co. m mam o. mN: ANI_V mc.vcmaw oxcnx mx oxcnx mx oxcnx mx co.umu:vu co.umu:vu _oozum co.umu:um A: u _cv mco_mcoe_a. m......8 ...... a... no... ...... N. 25.. ...... .... 5...... ”:22...“ m m N nu:c_ucouuuu_nm_.m> acovcmamvc. sumo Low mco.mcoe.u _m_00m tem u.EOcoum co mco.um_o..ou ma ms. mc_.maeounumum0u : >0:u.£3uccmzuu.m x_ozumm< 27I oN.. u ..o on>.o .. moo. n .o cn>.o .. moo. n ..=,cn>.o .. N u .o o u o o u .o oN m: .N w_xcnx NI m.N _NI .NN MI moo .o. co..nuonn 0. n.< . .No o. mom . oMM ... n..o.. o.onz N mm. m mm. m mNo N_Imv co.uuuuo.m .053mcou o NmN m xmo o NNN .o. n.n..nz .n.uom m o.M m M.. m wwfl .o. .oosn>o.oe. ono.o m NxN N MN. : _wmm_ ... mo...no.. ..>.o Am u ch mco.mcoe_o. >o__om _m_u0m NN o. :M w_xonx MI. CNN _MI mmm NI. Add Amn_v mc.vcmam woman 0 me N mam m MWM. N:n_v unmv cu co.umuco..o o. N.m m MNm. w mwm. A.I.v mmuc.m:n .m> .onm. N .mN . «ma. o. mwm AN-.. mc.oonom xonx mx oxonx mx axonx mx vo>o_aeoc: N 0800:. cm_uoz co_umu:no cm.umz A: u _cv mco_mcoE.o N. __ o. >u._om u.Eocoum to:c_ucouuno_nm..m> ucmvcoaovc_ :umo .ON mco_mcoe_v _m_00m tam 0.50:000 co mco.um_u..ou mm 0:. mc..maeouaumumou 3 >ocu_czuccwznu.u x_ozmmm< 272 moo. u ..o 83.. .. moo. u .o.oo>.m .. o.o. u ..o.cn>.o .. o u .o o a o . n ... . _N m: NN w xcmx NI MNo mI ..NN N. «mm 8. 833....n 3 n... _ mNN o. Nm: _ _mo Auv muzm.. 0.moz_ m WNW m mNN m NMM A.Imv c0.u00u0.a .meamcou .. ...I.IN m moo .. .oNN. .3 23.2. .223 m ”mm. m mm. m 0 _ Nov uc080>0.ae. cmn.= M N.N N ..N. N «mm N... 83.2... :3. No u ch mc0_mcoe_o >0__0m _m_00m :m 0_ mm N.xcm¢ NI a A ..o NI a .M-.. 9.38% 8an N mum N a o .|...m ...-.. 38.. 3 8338.5 m mm“. m WWW m uNdfl N_I_V mmoc.m:n .m> .0nm; o. mam . ooo o. MNM .N-: 9.38% nxcmm mm nxcmm mm nxcmm mm u.c: mc.._03u mc_nE:_a ..m nu.3 mu_c: mc.__0zv A: n _cv mc0.mcue.o .ma meoo. cm_voz m— 3 ...... venom oNo :— m. um.a:000 .oczo N >0._0m 0.80:00m . uo:c_uc0uuuo_nm..m> ucovcmaovc_ £000 .0» mc0.mcue_n .m.00m vcm 0.50:000 c0 m:0.um_o..0u mm 0;. mc..maeouunmummu 2 >0cu.:3IccszI.m x_ozmmm< 273 moo. u .o co>.o .. moo. u .o 8).m .. o.o. u .o on>.m .. o n 2 o u 2 _ u a m: m: a: w xcnx NI Nm. NI. NN. NN. NNN .o. co..nuoon o. n.< o. mN: o. mo: m mmw .m. muxo.. o.onz w mm. m mmN : mwm. A_Imv c0.u00u0.m .mesmcou m mNo m mo. N mmm. va 0.0».03 _m_00m o moo o Nm. m mmm. Nov acoeo>0.ae_ can.: N .m. w ooN w w . ... mu.u.un..,..>.0 Am a ch mc0.mcoe.a. >0..0m _m_00m o. o. ._ w xcmx ml. uNmm. MI. :00 MI. mmm Amu_v mc_ncoam 000nm N NNN M umNm . :No .3-.. .onn o. c0338.8 : .NNm N uNNm m Nam ..-.. mano.moo .m> .oan. . NNN . «mm M NNN .N-.. oo.nonom oxonx mx nxonx mx xcnx mx 3:0. cm.voz 0:.m> 0&0; cm.vux u.c: mc.__03u .: u .:. mc0_mcoe_o m. N. .ma m:0m.0a cm_voz >0._0m 0.50:00m m. voac_ucouIIo_nm_.m> ucmvcoaov:_ sumo .0» mc0_mcme_u _m_00m ucm 0.50:000 c0 mc0.um_0..00 mm ocu mc_.maeoounmummu 3 >0:u_:3IccszI.m x_ozmmm< 27h ..NN.-m.. .oo .omm.. .non.m nnmn .0a>u acmzuo.» 0.05 0;» 30.0n voxcm. m. 0a>u u:0:¢0.» mm0. ecu ..3000 muco_0_»»000 mm o>.ummoc ncm 0>_u_m0a »0 0.:ux_s m c0533 .c0.umu:aeou mum—a .30» m 50.» vocm_mmm xcmzu moo. u .o.on>.o .. m.o. u ..o 8>.m .. moo. n .o on>.o .. 0 fl 3 N nu .3 O nu 3 m: mN m: w.xcm¢ NI mmmfl Mw. NN. NI. Nm. .o. co..nu=nn o. n.< 0. we: _ mmo o. mJM Amv mucm_. 0.m02 m ANN m m:— w .m. A_Imv c0.uuou0.m .meamcou m NNN. m .o. m moo .o. n.n..nz .n.oom o NNN. o w.N m .o. .o. .cnsn>o.oe. cnn.o m NNN : mm. N m:. ... mn.3.no.. ..>.o .m n ch mc0_mcoe.o. >u..o. .n.uom o. NM o. w xcnz MI mum. mfl JMN MI. ”mm. AMI_V mc.vcuam oomam N MMN m UNON m m_o A:I_v “new on c0_umuc0..o M mmw m MMN N NNN. ..-_. mmno.no. .m> .oonx . moN N UNoN . NNm .N-.. oo.nonom xcnx ax xcnx mx oxonx ax .oe.mu x m.0x.03 m.0x.03 A: u _cv mc0_mcme.a .m.—00 0:_m N .m.—00 0a.;3 X >0__0¢ 0.50:00m ON m— um:c_ucooano_nm_.m> ucmucoaouc_ £000 .0» mc0_mcme_v _m_00m new 0.50:000 c0 mc0.um_0..00 mm on. mc_.m0500IImummu 2 >0:u.:3uccszI.m x_ozmmm< 275 mod wfl a Nola mum «N flw molo. BEE .\. ..N muo ouo NN. ___ moo ... omo moo mouxooz Lo:oU oo_o.x .oN NNN No: omN on: mmm oo: NNN om: mooxooz .m__ou oo_;: x .m_ mom mom :om mmm mm: mom NNN moo you; cm_ooz .m_ mom oo: Nmm om: “on oom Nm. m_o m=_m> «so; c._o.z, .N_ mmd. NflM. wad. MMN mad. mmfl. mam ANN. u_c: mc___03u Lug mcomgma cm_umz .m. «mm «No mmo omo «mm. ouo mam. mmo o_co oc___ozo Loo meooo cm_oo: .m. mmm. mwm. Jon mmm mo: Nmm NMN m_o ocono_o :m so“: mo_co ocoom x .:. «_o Noo “mm. wNm «ma. «mm. mwm .mma mu_co oo___ozo oo_oouuo Luczoox .m. _oN omm moo __N mNN mom om. .NN oo>o_o50co x .No _om omm :o: _om mo: ouo sou _No ueouc_ cm_ooz .__ :MN mom ooN “on _mo own .m. _m: co_omuoou cm_ooz .o. omo mo. muo o:. m__ mo. mmo mm. comomoooo oou__ou ;o_: x, .m mmm. :_m mo. mmm. _:~ mmm so. mom co_omuoom .0050. ;o_; ;o_3_x .o moo «mm. mam: zoo Nww mNm «ma Nam. co_omuooo 3o. ;o_: x. .N oo: mom _m: mum om: :mm mom Nmo co_omuoou xomoomeo_m oom>_oo ;u_: x .o.~ MNN om: o_m mo: _om _mm mm. KN: mom cm_ooz .m 8: E o? Mono a mo an m; .38... area a .o «NM .3 mod a mum «mm v.3 .ola. o.o.: x .m mmm om: mmm. mmm_ mNJ Rom oNN Koo smog: x, .N _No “mm. __o «:o «mm. mam_ mmm mmm. uocmgu co_om_oooo x ._ _I. d S J... 3 n S 3 E E 0 O O J .0 I. 0.. J 3 J U a. a A O 1 I. a S p. U I. J 4A al.. ..m. m U W I. A I. U A )I. U I.. )5 O M p. a w 6 I. I . IA 8 d 1. (d O. ....0. oh H m. m. m \U m 332...; acmucoaovf (.m. m o w .... m. U \.I a 9 S a mm Mu mu L. w x; S I.\ ..l. ( NW >_co mumouoemonumo_nm_gm> mc_uo> __mu __0L 0. van mm_nm_om> >uc0:u_umcou ucmucoaonc_ _N cmozumn mco_um_oLLou och-n.u x_ozumm< 276 whom mm a a H _Iod «Mm MNM 55...... .N. ..N ooo ooo o2 NN. o3 mu. :mo 3o 29:2 3:8 3; .m. .8 So So oom 2m Rm oo: o; «S 29.32 L.....__8 32: .m .m. flm omm a: ooo mom moo m3 9: 28 ago»: .2 law mmu Nom .mlom. 3m mmm ...mh omo 2:? 25o :28: .2 mmm. mam. mum mm. NMfl NWN. mm. :mw. u_c: mc___03c ova mCOmooa cm_um: .m— H ooo mmo No to mo. mum ooo :5 852.6 .2. £08 :23: .m. Noo 2: o? oom _Iom Nmm HG Rb. 33an :m 53 3:5 .258 .m 4: moo ooo mom moo ooo a ooo oo_ 3:5 82.0% 8338 35.5 .N. .2 mm. mmN om. moo o: m3 «om omo 85352: .N. .N. _oo ~_: mm: oo: o_: moo mm: mm. «souc_ cm_ooz ... m_o omo mmm omw omu m_: loom ooo co_omuooo cm_ooz .o_ .8 .:o moo om. o3 m3 moN a 5:836 ooo:8 £5 .m .m Ro E im Km to 3m mmm ooo 8:82.» .85.. cm; 55 .m .o filo. om: a ooN a Nfl .odw mum 538%» :3 55 .m .m a mm: Sm Rm mm: omo mm or 5:838 €355... 39:3 53 .m .o mom .3 SN o2 NoN o3 Now :9. mm... :28: .m olmm «om mum _mm mu: mom oom .m..m.m 18:33.8 o. .o a wow im _Imw HM oIom wow so Emozo. .m .6 _mm mmm mom in Em :3 No. 535$ .N _No mod o8 m8 omo 3o 2. a omcmou 8:238 .m ._ I: V d 0 S 0 N 3 V 0 I. J J .0 J a O 5 J p O I. e I. 5 U J a ...: a 3 a J S I. I. .4 9 U a U 0 n O 5 O S 3 .4 w n U \II S D. )S )9 J )3 II 9 8 I. .4 Id I 14 I. 8 J )1. 1. (p O I. . a - I. 6 - 3 n u. m w m Mow. Wm H Um (m 822.2, 28:33:. D. p. I I. S O I 9 mule. a“. m 9W m \.I 13w .0 W w m p G n m w. .H; w m. m. ( ( 14 U :A mc_uo> __mu __0; m_ ucm mm_nm_om> >ucm:u_umcou ucmucmamvc_ _N comzumn mco_um_mgoou to::_ucouuu>_co mumouoeoonumu_nm_gm> Ocoun.u x_ozmmm< 277 NM NS. m @ oo. mm oo. 3 828“. .m .3 mo. «mm :o ooo a ooo omo moo 89.82 8:8 8:. .:. .oo mm mm. .o. m; :oo o: mo. 3. 29.82 8:8 8:: .m .m_ ooo m: ooo moo MINIM of mo io 28 8:8: .o. ooo mo_ om. oom ooo moo ___ mo_ oo_m> 820: om_o8: .N. No. moo mo. 2. a :1 @ mod :8 8:88 8o .838 888: .o. om. oo. :9 N: a Nmo o_o omo :8 8:88 .8 888 88: .m. mmo mo. .mli. mmo mlo. omo oo_ omo 8383 :8 £3 3:5 .088 .N. .i «.8 2o :: @ oo. a 9|. 3|. 3:5 8:88 8888 8:so .:. .m. om. a omo No. mmo to ooo So 8338:: .:. .N. mmo. o: om. mom Nolo ZN .o. mo. 2.85 8:8: .: __o mo. moo ~_~ o:_ _oo _oo ooo co_omuoo8 o8_o8: .o. mmo of mm. {R .8 mo. mmo om. 8:828 om8:8 85 .m .m a mo. mmo ooo Won. .8 omo ooo 8:828 .88. 8E 55 .:. .o omo 3o moo 2o Noo om. E mo 8:828 :3 £5 om .5 oo. ooo NR .:m a omo No. a: 8:828 3388.8 88:8 £5 .:. .o o_m oo_ ~_~ mm. moo mom m:_ ooo 8mm cm_ooz .m 2o Wow oom ooo m: o? Rm om. V88... 888 .:. .: go .3 2o .8 a ooo ooo :mo 9.82 .m .m o: :2 m.o|_ Now mollo moo mmo MIC 88o .:. .N am to ooo moo :2 Na wolo ooo 888 82838.. .:. ._ .I d S 4. 3 n S 3 9 E O O O J d I. O. J 3 J U 0.. 9 A o 1 I. a S e u I. J IA D. I. 9 U D- I I 6 J I. A I. U A )I. U I )S O M 9 am 6 I. I . :A 3 d 1. (d O. h. m. I. m. I. J nu H mm m_:m> ucmucoaouc_ (m. .4 me: I. w m . 1 _a . S :A J D 3 Z I. I. a :A \.I w (\ a U ) a 9 S o m. m, W n. w ) S ( ...I (\ I.” >_co mcmu__n:aomnumo_nm_om> mc_uo> __mu __oL m. ucm mm_nm_om> >uco:u_umcou acmccmaouc_ _N comzumn mco_um_mooou o::::.u x_ozmmm< 278 ~_o I_.m.~. an «mm NH mom a m? 858.. .m .8 omo m_o omo :o ooo a ooo Mom 89:03 8:8 83;. .:. .o~ % moo ooo Nom om_ om_ o3 an 29:03 8:8 823 o. .m. 3o o: :8 omo ooo oo. mmo odd 82 888: .o. ooo mm. _mm in ooo moo mom m 88> 88 8:8: .: ooo io 3o m: 3. N: o: E :8 8:88 88 8888 8:8: .o. mam om_ o8 omo ooo m_m Bo I~|_ :8 8:88 .88 88.. 888: .m. :mo .3. 2m .oom :2 3a m_m mIoI 8388 _8 £3 8:5 838 .m 4: mom m_o 94 .m filo .on ._Imb oi 8:5 8:88 .8888 88o .N. .9 ME 3o ooo 3o ooo :o. .:o ooo 8888c: .:. .N. Nmo ooo mom m_m mi o_m mom Nm. 8585 8:8: .: mko. m: om_ _oo omo +a. N: mod 8:888 888: .o. .:o moo omo mom mm. m: moo mlmlw 8:888 88:8 85 ..m .m m_o oo. o: _mN Molo :: Nm. _oo 8:888 .88. 82 55 .N. .o Rm omo moo NN. o_o ado. ooo molo 8:828 8. 85 .m .: :oo oo. mmo m:. m__ ooo ooo ooo co_omoooo :omocoeo_8 oom>_.o :o_: x .o ”AIM oo. omm om_ mmo omo mi a 8mm 8:8: .m ooo m2 _Nm mmm _: omo mom ooo :88 888“. .m .: mmm So «mm N: o_o ooo m: H 882 .m .m Mm mom ooo No: i. mo. ooo MW 88: o. .N ooo 2o moo So no _oo :2 _ 888 8:888“. .m ._ I: V d 0 S 0 N 3 V O I. J J .0 J 8 O 5 J p O I. B I. 6 U J 9 I38 3 9 J S I. I. 1. 9 U 9 U 0 n 3 5 O S 1. 1. w n U ) S 9 \.IS )9 J )9 , I 9 9 I. .4 Id I 1. I. 9 J )1. 3 (\P O I. . a . I. 5 . 3 n m. m whnuv “um.” 1....wa m. hum (m. om_nm_._m> ucmncoamvf a n. mm. m m ) a d W m w p G n m. . ) a I. I. 7o H q 0 D II\ ( 14 U [A mc_uo> __mu __0: m. vcm om_nm_:m> >ucm:u_umcou acoucmamnc_ _N cmmzumn mco_um_m::ou um:c_ucouuu>_co mcmo__n=ammnumw_nm_om> OLLII.u x_ozmmm< BIBLIOGRAPHY 279 BIBLIOGRAPHY Adrian, Charles R. Governinqurban America (New York: McGraw-Hill Book Company, I961). Adrian, Charles R. and Press, Charles. The American Political Process (New York: McGraw-Hill Book Company, I965). Barbash, Jack. The Practice of Unionism (New York: Harper and Brothers, I956). Beard, Charles A., and Lewis, John D. ”Representative Government in Evolution” American Political Science Review XXVI (April, I932) pp. 223-2&0. Becker, Robert W., Foote, Frieda L., Lubega, Mathias, and Monsma, Stephen V. ”Correlates of Legislative Voting: Michigan House of Representatives, l95h-I96l” Midwest Journal of Political Science VI, No. 4 (November, I962) pp. 384-396. Belknap, George M. ”A Method for Analysing Legislative Behavior“ Midwest Journal of Political Science || (I958) pp. 377-“02. Berelson, Bernard R., Lazarsfeld, Paul F., and McPhee, William N. Voting (Chicago: University of Chicago Press, I954). Berman, Daniel M. A Bill Becomes a Law: The Civil Rights Act of [960 (New York: The MacMilIan Company, I962). Burke, Edmund. Reflections on the Revolution in France (Chicago: Gate- way Editions, lnc., Henry Regnery Company, originally published in I790, republished in I955). ' Campbell,.Angus., Converse, Philip E., Miller, Warren E., and Stokes, Donald E. The American Voter (New York: John Wiley and Sons, Inc., l960). Campbell,.Angus, Gurin, Gerald, and Miller, Warren E. The Voter Decides (New York: Harper, Row, and Company, I95h). Cantril, Hadley. (ed.), Public Opinion, l935-l9h6 (Princeton: Princeton University_Press, I951). 280 28] Cnudde, Charles F., and McCrone, Donald J. ”The Linkage between Con- stituency Attitudes and Congressional Voting Behavior: A Causal Model” American Political Science Review LX, No. l, (March, I966), pp. 66-72. Committee on Political Parties of the American Political Science Association, ”Toward a More Responsible Two-Party System” American Political Science Review XLIV (September, I950) Supplement. Congressional Quarterly Service, CQ Census Analysis: Congressional Districts of the United States (Washington: Congressional Quarterly Inc., l96h). Cornwell, Elmer E., Jr. ”Bosses, Machines, and Ethnic Groups” The Annal of the American Academy of Political and Social Science CCCLIII (May, 1964), pp. 27-39. Cox, E. F. ”Congressional District Party Strengths and the 1060 Elec- tion” Journal of Politics XXIV (I962), pp. 277-302. >Crane, Wilder Jr. 8A Caveat on Roll-Call Studies of Party Voting” Midwest Journal of Political Science IV, (I960), pp. 237-249. Cutright, Phillips. ”Political Structure, Economic Development, and National Social Security Programs” American Journal of Socioloqy Lxx (I965). pp. 537-550. Dahl, Robert A. Conqress and Foreign Policy (New York: W. W. Norton and Company, Inc., I96h; first published by Harcourt, Brace and Company, Inc., I950). . Pluralist Democracy in the United States: Conflict and Con- sent (Chicago: .Rand McNally and Company, I967). Dawson, Richard E., and Robinson, James A. ”Inter-Party Competition, Economic Variables, and Welfare Policies in the American States” Journal of Politics xxv (I963), pp. 265-289. Dawson, Richard E., and Robinson, James A. ”Politics of Welfare” in Herbert Jacob and Kenneth N. Vines (eds.), Politics in the Ameri- can States: A Comparative Analysis (Boston: Little, Brown and Company, I965), pp. 37l-4I0. deGrazia, Alfred. Public and Republic: Political Representation in America (New York: .Alfred A. Knopf, I951). See especially Chapter II, ”English Ideas in Colonial Times,’l pp. l3-49. Dexter, Lewis A. I'The Representative and His District” Human 0rqaniza- tion XVI (I957), pp. 2-l3. 282 Dixon, W. J. (ed.) BMD: Biomedical Computer Programs (Los Angeles: Health Services Computing Facility, Department of Preventive Medicine and Public Health, School of Medicine, University of California, Los Angeles. January I, I96h; Revised September I, 1965) . Dye, Thomas R. Politics Economics and the Public: Policy Outcomes in the American States (Chicago: Rand McNaIly and Company, I966). . ”State Legislative Politics,” Chapter 5 in Herbert Jacobs and Kenneth N. Vines, Politics in the American States: A Comparative Analysis (Boston and Toronto: Little, Brown, and Company, I965), pp. 151-206. East, John Porter. Council-Manager Government: The Political Thought of Its Founder, Richard S. Childs (Chapel Hill: The University of North Carolina Press, I965). Easton, David. A Framework for Political Analysis (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., I965a). A Systems Analysis of Political Life (New York: JohnIWiIey and Sons, Inc., l965b). Ennis, Philip H. ”The Contextual Dimension in Voting in William N. McPhee and William A. Glaser (eds.) Public Opinion and Conqres- sional Elections (Glencoe: The Free Press of Glencoe, I962), pp. I80-2Il. Farris, Charles D. ”A Method of Determining Ideological Groupings in Congress,” The Journal of Politics XX (I958), pp. 308-338. Reprinted in John C. Wahlke and Heinze Eulau, Legislative Behavior: A Reader in Theory and Research (Glencoe: The Free Press of Glencoe, I959). Fenno, Richard F., Jr. ”The House Appropriations Committee as a Politi- cal System: The Problem of Integration” American Political Science Review LVI (June, I962), pp. BIO-324. Flinn, Thomas A. ”Party Responsibility in the States: Some Causal Fac- tors” American Political Science Review LVIII, No. I, (March, 196%). pp. 60-71. Frost, Murray. ”Roll Call Cohesion in the New Jersey Legislature, I956- I957” (unpublished M.A. Thesis, Department of Political Science, Michigan State University, I959). Froman, Lewis A., Jr. Congressmen and Their Constituencies (Chicago: Rand McNally and Company, I963). 283 Froman, Lewis A., Jr., and Ripley, Randall 8. ”Conditions for Party Leadership: The Case of the House Democrats'I American Politi- cal Science Review LIX (March, I965), pp. 52-63. Garceau, Oliver. ”Research in the Political Process” American Political Science Review XLV, No. I, (March, 195]), pp. 69-85. Goldman, Ralph M. The Democratic Party in American Politics (New York: The MacMilIan Company, I966). - Gray, Charles H. ”A Scale Analysis of the Voting Records of Senators Kennedy, Johnson, and Goldwater, l957-I960'I American Political Science Review LIX, No. 3 (September, I965), pp. 6l5-62l. Greenstein, Fred I., and Jackson, Elton F. ”A Second Look at the Validity of Roll-Call Analysis Midwest Journal of Political Science VII, No. 2 (May, I963). PP. l56-I66. Grimes, Alan.P. Equality in America: Religion, Race, and the Urban Maiority (New York: Oxford University Press, I96h). Gross, Bertram M. The Legislative Struggle: A Study in Social Combat (New York: McGraw-Hill Book Company, Inc., I953). Guilford, J. P. Psychometric Methods (New York: McGraw-Hill Book Com- pany, Inc., l95h). Hadden, Jeffrey-K., and Borgatta, Edgar F. American Cities: Their Social Characteristics (Chicago: Rand McNally and Company, I965). Hays, William L. Statistics for Psychologists (Neszork: Holt, Rine- hart and Winston, I963). Huntington, Samuel P. ”A Revised Theory of American.Party Politics” American Political Science Review XLIV, No. 3, (September, I950) pp. 669-677- Jacob, Herbert. ”State Political Systems” in Herbert Jacob and Kenneth N. Vines (eds.), Politics in the American States: A Comparative Anal sis (Boston and Toronto: Little, Brown, and Company, 1965 , pp. 3-21. Jewell, Malcolm E. ”Party Voting in American State Legislatures” American Political Science Review XLIX (I955), pp. 773-79I. . The State Legislature: Politics and Practice (New York: Random House, I962). Jones, Charles 0. ”Representation in Congress: ‘The Case of the House Agriculture Committee“ American Political Science Review LV (June. 1961). PP- 358-367. 28h Jones, Charles 0. The Republican Party in American Politics (New York: The Macmillan Company, I965). Keefe, William J. and Ogul, Morris S. The American Legislative Process: Congress and the States (Englewood Cliffs: Prentice-Hall, Inc., 1963). Key. V. 0., Jr. American State Politics: An Introduction (New York: Alfred A. Knopf, I956). . Politics, Parties, and Pressure Groups 5th ed. (New York: Thomas Y. Crowell Company, l96h). fgblic Opinion and American Democracy (New York: Alfred A. Knopf, l96l). . Southern Politics in State and Nation (New York: Alfred A. Knopf, Inc., l9h9). Key. V. 0., Jr., and Munger, Frank. ”Social Determinism and Electoral Decision: The Case of Indiana“ in Eugene Burdick and Arthur J. Brodbeck (eds.), Americgn Voting Behavior (Glencoe: The Free Press, l959), pp. 28I-299. Krislov, Samuel. ”What is an Interest? The Rival Answers of Bentley, Pound, and Maclver” The Western Political Quarterly XVI, No. A, (December, I963), pp. 830-h3. Lane, Robert E. Political Life: Why People Get Involved in Politics (Glencoe: The Free Press, l959). Lazarsfeld, Paul F., Berelson, Bernard R., and Gaudet, Hazel. Ihg. People's Choice (New York: Columbia University Press, I9hh and l9u8T. Lowell, Lawrence A. ”The Influence of Party Upon Legislation in England and America” Annual Report of the American Historical Associa- tion for I90I, I, pp. 32l-5hh. MacRae, Duncan ”A Method for Identifying Issues and Factions from Legis- lative Votes” American Political Science Review LIX (December, 1965). pp. 909-926. . Book Note on Froman, Congressmen and Their Constituencies, in The American Political Science Review LVIII (June, l96h), pp. #95- #96. . Dimensions of Congressional Voting: A Statistical Study of the Hggse of Representatives in the Eighty-first Congress (Berkeley and Los Angeles: University of California Press, I958), pp. 203-390. 285 MacRae, Duncan, Jr. “Occupations and the Congressional Vote, I9h0- I950” American Sociological Review XX, No. 3, (June, I955), pp. 332-350. . ”Some Underlying Variables in Legislative Roll Call Votes” Public Qpinion Quarterly XVIII, No. 2, (I959), pp. I9I-l96. ”The Relation Between Roll Call Votes and Constituencies” American Political Science Review XLVI (I952), pp. l046-I055. Marx, Fritz Morstein. ”Party Responsibility and Legislative PrOgram” in John C. Wahlke and Heing Eulau (eds.), Legislative Behavior: A Reader in Theory and Research.(GIencoe: The Free Press of Glencoe, Illinois, I959), pp. 55-59. First published in Columbia Law Review L, (March, I950), pp. 28l-99. Matthews, Donald R. U. S. Senators and Their World (New York: Vintage Books, A Division of Random House, I960). Mayhew, David R. Party Loyalty Among Congressmen: The Difference Between Democrats and Republicans, I9h7-I962 (Cambridge, Massachusetts: Harvard University Press, I966). McNemar, Quinn. Psychological Statistics 3rd ed. (New York: John Wiley and Sons, Inc., I962). McQuitty, Louis L. ”Elementary Linkage Analysis for Isolating Ortho- gonal and Oblique Types and Typal Relevancies” Educational and Psychological Measurement XVII (I957) pp. 207-229. Menzel, Herbert. ”A New Coefficient for Scalogram Analysis” Public Qpinion Quarterly XVII (I953), pp. 268-280. Meyer, John W. “A Reformulation of the 'Coattails' Problem” in William N. McPhee and William A. Glaser (eds.) Publichpinion and Congressional Elections (Glencoe: The Free Press of Glencoe, I962). pp. 52-6h. Miller, Warren E., and Stokes, Donald E. ”Constituency Influence in Congress“ American Politiggl Science Review LVII (March, I963), pp- 45-56- Mitchell, William C., The American Polity: A Social and Cultural Inter- pretation (New York: The Free Press of Glencoe, A Division of The Macmillan Company, I962). Monsma, Steven V. I'Informal Groups in the Legislative Process: A Study of the Michigan House of Representatives” (Unpublished Ph.D dissertation, Dept. of Political Science, Michigan State Univer- sity, I965). 286 Monsen, R. Joseph, Jr., and Cannon, Mark W. The Makers of Public Policy: American Power Groups and Their Ideol0gies (New York: McGraw Hill Book Company, I965). Moos, Malcolm. Politics Presidents and Coattails (Baltimore: The Johns Hopkins Press, I952). Morris, John. Technical Report 4;, Nonparametric Chie§guare Tests and Agelyses of Variance (East Lansing: Michigan State University, Computer Institute for Social Science Research, June I, I966). Potter, David M. People of Plenty:. Economic Abundance and the American Character (Chicago: University of Chicago Press, I95h). Press, Charles. “Voting Statistics and Presidential Coattails” American Politicgl Science Review LII (I958), pp. IOAl-IOSO. Press, Charles, and Adrian, Charles R. ”Why Our State Governments are Sick” Antioch Review XXIV (Summer, I964), pp. Ih9-l65. Ranney, Austin, and Kendall,.Willmoore. Democracy and The American Party System (New York: Harcourt, Brace and Company, I956). Redford, Emmitte S. Ideal and Practice in Public Administration (Bir- mingham: University of Alabama Press, I958). Riedel, James A. ”Boss anleaction” The Annals of the American Academy of Political and Social Science CCCLIII (May, I96h), pp. lh-26. Reiselbach, Leroy N. ”The Demography of the Congressional Vote on Foreign Aid, I939-I958” American Political Science Review LVIII, No. 3, (September, lgeu), pp. 577-588. ”Foreign Policy Ideology in the 88th Congress: Constituency and Other Correlates” Unpublished;paper presented to the I966 Annual Meeting of the Midwest Conference of Political Scientists, Chicago, Illinois, April 28-30,.I966. Robinson, James A. The House Rules Committee (Indianapolis and New York: The Bobb's-Merrill Company, Inc., I963). Robinson, W. S. ”Ecological Correlations and the Behavior of Individuals” Americgn Sociological Review XV (I950), pp. 35I-357. Rokeach, Milton. ”The Nature of Attitudes” (Mimeographed, pp. l8. To be published in the InternationeI_Encychpedia of the Social Sciences). Schattschneider, E. E. Eerty Government (New York: Rinehart and Com- pany, Inc., I9h2). Schlesinger, Joseph A., Ambitionggnd Politics: Political Careers in the United States (Chicago: Rand McNally and Company, I966). 287 Schlesinger, Joseph A. .”A Two-Dimensional Scheme for Classifying the States According to Degree of Interparty Competition” American Political Science Review LIX (I955), pp. Il20-II28. . .“Political Party Organization” in James G. March (ed.), fiendbook of Organizations (Chicago: Rand McNally and Company, I965). pp. 76h-801. Schubert, Glendon, ”Academic Ideology and the Study of Ad'udication” Americgn Political Science Review LXI (March, l967l, pp. IO6-l29. . Quantitative Analysis of Judicial Behavior (Glencoe, Illinois: The Free Press, I959) Chapter V, ”Scalogram Analysis,” pp. 269- 376, was useful for my purposes. Siegel, Sidney. Nonparemetric Statistics for the Behavioral Sciences (New York: McGraw-Hill Book Company, Inc., I956). Snowiss, Leo M. ”Congressional Recruitment and Representation” American Political Science Review LX, No. 3, (September, I966), pp. 627- 639. Sorauf, Frank Jr. Party and Representation: Legislative Politics in Pennsylvania (New.York: .Atherton Press, A division of Prentice- HaIl, Inc., I963). Spaeth, Harold J., and Peterson, David J. ”The Dimensionality of Civil Liberties Decosion Making: A Longitudinal Analysis” Mimeo- graphed paper, I967. Stokes, Donald E., and Miller, Warren E. ”Party Government and the Saliency of Congress” The Public Opinion Quarterly XXVI, No. 4, .(Winter, I962). pp. 53l-5h6. Stouffer, Samuel A., Guttman, Louis, Suchman, Edward A., Lazarsfeld, Paul F., Star, Shirley A., and Clauson, John A. Measurement and Prediction (Princeton: Princeton University Press, I950, and New York: John Wiley and Sons, Inc., I966). . Torgerson, Warren S. Theory and Methods of Scaling (New York: John Wiley and Sons, Inc., I958). Truman, David B. The Congressionelgferty: A Cese Study (New York: .John Wiley and Sons, Inc., I959). The Governmental Process: Politiggl Interests and Public Opinion (New York: .Alfred A. Knopf, Inc.,l95l). 288 Truman, David. ”The Presidency and Congressional Leadership: Some Notes on Our Changing Constitution” Proceeding; of the American Philosophical Society, October I959, pp. 687-692; reprinted by Donald Bruce Johnson and Jack L. Walker (eds.), The Dynamics of the American Presidency (New York: John Wiley and Sons, l96h), pp. 2II-2I6. Turner, Julius, Party and Constituency: Pressures on Congress (Balti- more: Johns Hopkins Press, l95l). . ”Primary Elections as the Alternative to Party Competition in 'Safe' Districts” Journal of Politics XV (I953), pp. I97-2IO. U. 5. Bureau of the Census, Congressional District Data Book (Districts of the 88th Congress A Statistical Abstract Supplement. U. S. Government Printing Office, Washington, D. C., I963. Wahlke, John C., Eulau, Heing, Buchanan, William, and Ferguson, Leroy C. The Legislgtive System: Explorations in Legislative Behavior (New York: John Wiley and Sons, Inc., 1962). Williams, Oliver P., and Adrian, Charles R. Four Cities (Philadelphia: University of Pennsylvania Press, I963 . Williams, Oliver P., Herman, Harold, Leibman, Charles S., and Dye, Thomas R. Suburban Differences and Metropolitan Policies: A Philgdelphia Story (Philadelphia: University of Philadelphia Press, I965yf Wilson, James Q. Negro Politics: The Search for LeaderShIQ (New York and London: The Free Press and Collier-Macmillan Limited, 1960 and 1965). Wilson, James Q., and Banfield, Edward C. ”Public-Regardingness as a Value Premise in Voting Behavior” American Political Science Review LVIII (December, l96h), pp. 876-887. Wolfinger, Raymond E., and Heifetz, Joan. ”Safe Seats, Seniority, and Power in Congress” American Political Science Review LIX (June. 1965). pp. 337-359. Young, Roland. The American Congress (New York: Harper and Brothers, Publishers, l958). Zelditch, Morris, Jr. A Basic Course in Sociol0gical Statistics (New York: Holt, Rinehart and Winston, 1959). "IIIIIIIIIIIIIIIIIII'IIII'“ 3 77 3231